US20170091846A1

Movatterモバイル変換

Info

Publication number: US20170091846A1
Application number: US15/268,163
Authority: US
Inventors: Takuo HENMI; Shigefumi IINUMA; Takaaki OKADE
Original assignee: Universal Entertainment Corp
Current assignee: Universal Entertainment Corp
Priority date: 2015-09-25
Filing date: 2016-09-16
Publication date: 2017-03-30
Also published as: CN106991107B; CN106991107A; PH12016000332A1; JP6842825B2; US11798053B2; JP2017062741A; US20200320604A1

Abstract

Provided is an information provision system which determines preference of a user based on a dialogue with the user utilizing an Internet shopping site and determines a recommended commodity based on the determined preference. An information provision server in an information provision system receives utterance (input) by a user utilizing an Internet shopping site via a user terminal and provides a response thereto onto the user terminal, thereby performing control so as to have a conversation with the user. Further, based on the input by the user and user attributes, the information provision server determines a user type of the user and when a recommended commodity is presented to the user, determines the recommended commodity from among commodities purchased by other user whose user type is the same as the user type of the user.

Description

CROSS-REFERENCE TO RELATED APPLICATION

This application claims the benefit of Japanese Patent Application No. 2015-188769, filed on Sep. 25, 2015, which application is incorporated herein by reference in its entirety.

FIELD OF THE INVENTION

The present invention relates to an information provision system which provides information in response to a user input.

BACKGROUND OF THE INVENTION

In recent years, vigorous studies on a system which is operable to search an optimum solution to a question from a user and to respond thereto have been carried out. Further, in connection with the development of the above-mentioned system, studies on a natural language processing technology and an artificial intelligence technology for analyzing a syntax and meaning of a question inputted from a user and for searching a solution to the question have also been in progress.

For example, a question answering system which is operable to output answer candidates evaluated as optimum ones in response to a question has been proposed (refer to U.S. Patent Application Publication No. 2015/0026106). This question answering system is configured such that answer candidates to a question are searched; with respect to each combination of each of the answer candidates and the question, a score indicating a degree of each of the answer candidates being a correct answer to said question is calculated by using a learning function; and based on the calculated score, an answer candidate which is most likely to be a correct answer to the question is outputted.

In addition, a request content identification system which is operable to analyze an input sentence obtained by describing a request from a user in a natural language; to determine whether the request content of said input sentence corresponds to the request content previously defined; and to thereby identify and output the request content of the input sentence has been disclosed (refer to U.S. Patent Application Publication No. 2009/0099840).

On the other hand, on the existing Internet shopping sites, on each of which a user purchases commodities and the like via the Internet, a recommendation function with which when a user selects a desired commodity by a name or a category of a commodity, commodities which other user having purchased the selected commodity has purchased concurrently with purchasing the selected commodity and commodities which this other user has so far purchased are presented as “recommended commodities” has been provided.

However, although on the above-described existing Internet shopping sites, by using the recommendation function, the commodities purchased by other user can be presented as the recommended commodities to a utilizing user, in the relationship between the utilizing user and other user, only the point that the utilizing user and other user have purchased the common commodity (or the utilizing user and other user have examined purchasing of the common commodity) is shared by the utilizing user and other user, no information based on actual preference or attributes of the utilizing user is available, and any information related to optimum commodities based on preference of a utilizing user cannot be presented.

In addition, on the above-described existing Internet shopping sites, a function with which even in a case where a utilizing user inputs ambiguous search words, inference based on the search words is conducted, and commodities which are appropriate for the utilizing user are presented is not available. In addition, a function with which through a dialogue with a utilizing user, words for determining preference of the utilizing user are drawn out or a side of an Internet shopping site actively induces a utilizing user to purchase a specific commodity has also not been provided.

In the above-described question answering system disclosed in U.S. Patent Application Publication No. 2015/0026106 and the request content identification system disclosed in U.S. Patent Application Publication No. 2009/0099840, in response to a user's input, answer candidates can be presented. However, the mere combination of these systems and the system of the existing Internet shopping sites does not allow preference and the like of a utilizing user to be grasped through a dialogue with said utilizing user doing Internet shopping. As a result, a function to present appropriate recommended commodities to said utilizing user cannot be realized.

Therefore, an object of the present invention is to provide an information provision system which is operable to determine preference of a utilizing user through a dialogue with the user shopping on an Internet shopping site and based on the determined preference, to determine a recommended commodity.

BRIEF SUMMARY OF THE INVENTION

The present invention provides an information provision system, an information provision method, and a storage medium described below.

A first aspect according to the present invention is an information provision system (for example, an information provision system1) including:

an input reception device (for example, aninput reception part201 in aninformation provision server200 shown inFIG. 7) for receiving an input (for example, character string data obtained from a character string and voice inputted by auser10 from a user terminal100) by a user from a user terminal (for example, the user terminal100) operated by the user (for example, the user10);

a response determination device (for example, aresponse determination part203C included in aresponse control part203 in theinformation provision server200 shown inFIG. 7) for determining contents of a response (for example, a response sentence and a sound effect output instruction transmitted from theinformation provision server200 to the user terminal100) to be outputted to the user terminal in response to the input; and

a user type determination device (for example, a recommendation relateddata updating part203B included in theresponse control part203 in theinformation provision server200 shown inFIG. 7) for determining a user type of the user (for example, user types A to X shown inFIG. 13) by using words (for example, key words associated with picky-about items shown inFIG. 12) obtained based on the input,

the response determination device determining a commodity or a service to be presented to the user based on the determined user type of the user,

the response determination device controlling information (for example, a commodity name, image data, detailed information (a URL), and the like) pertinent to the determined commodity or service so as to be included in the contents of the response (for example, such that such information pertinent to the commodity, included in the contents of the response, is displayed as a list of recommended commodities in acommodity display part126 shown inFIG. 3).

By employing the above-described configuration of the present invention, through a dialogue with a user utilizing an Internet shopping site, preference of the user is determined and based on the determined preference, recommended commodities to be presented to the user are determined. Therefore, the preference of the user is grasped in a specific and detailed manner and commodities which are appropriate for the user are presented as the recommended commodities.

In the first aspect, a second aspect according to the present invention is configured such that based on the determined user type of the user and a category (for example, a category selected by the user on a screen of an Internet shopping site displayed on theuser terminal100 and a category of the displayed and selected commodity) selected by the user on the user terminal, the response determination device determines the commodity or the service to be presented to the user.

By employing the above-described configuration of the present invention, in the form in which the category of the commodity or the like in which the user utilizing the Internet shopping site is currently interested is reflected, recommended commodities are determined. Therefore, the preference of the user and the category in which the user is currently interested are grasped in a specific and detailed manner, and commodities which are appropriate for the user are presented as the recommended commodities.

In the first aspect, a third aspect according to the present invention is configured such that based on purchase information (for example, information pertinent to the purchase or the reservation of a commodity or a service such as purchase results grasped from purchase history data (including ranking or the like related to purchased commodities in a predetermined period of time), purchase examining commodities, and commodity browse history) related to a user whose user type is the same as or similar to the determined user type of the user (for example, having predetermined communality with the user type of the user or belonging to the same superordinate classification), the response determination device determines the commodity or the service to be presented to the user.

By employing the above-described configuration of the present invention, based on the commodities purchased by other user whose user type is the same as or similar to the user type of the user utilizing the Internet shopping site, the recommended commodities are determined. Therefore, the commodities or the like purchased by other user having the preference in common with this user are grasped in a specific and detailed manner, and commodities which are appropriate for the user are presented as the recommended commodities.

In the first aspect, a fourth aspect according to the present invention is configured such that based on the words obtained based on the input and attributes of the user (for example, sex and age of the user), the user type determination device determines the user type of the user.

By employing the above-described configuration of the present invention, a user type of a user having the same attributes as those of the user utilizing the Internet shopping site is set, and based on commonality in such user attributes, recommended commodities are determined. Therefore, the preference of the user and the attributes of the user are grasped in a specific and detailed manner, and commodities which are appropriate for the user are presented as the recommended commodities.

In the first aspect, a fifth aspect according to the present invention is configured such that in accordance with the words received over time, the user type determination device updates the user type of the user (for example, after the lapse of a predetermined period of time, when predetermined words inputted by the user are often extracted and the preference or the picky-about items of the user seem to change to others, in addition to the current user type (or instead of the current user type), a new user type is set).

By employing the above-described configuration of the present invention, when the preference of the user utilizing the Internet shopping site has changed, based on the user type for which the changed preference is reflected, recommended commodities are determined. Therefore, in accordance with the change in the preference of the user according to the lapse of time, said preference is grasped in a specific and detailed manner, and commodities which are appropriate for the user are presented as the recommended commodities on an as needed basis.

A sixth aspect according to the present invention is an information provision method implemented on an information provision server in an information provision system, the method including:

an input reception step of receiving an input by a user from a user terminal operated by the user;

a response determination step of determining contents of a response to be outputted to the user terminal in response to the input; and

a user type determination step of determining a user type of the user by using words obtained based on the input, wherein at the response determination step, based on the determined user type of the user, the commodity or the service to be presented to the user is determined, and information pertinent to the determined commodity or service is controlled so as to be included in the contents of the response.

A seventh aspect according to the present invention is a non-transitory computer-readable storage medium having stored therein a program for causing a computer to function as:

an input reception device for receiving an input by a user from a user terminal operated by the user;

a response determination device for determining contents of a response to be outputted to the user terminal in response to the input; and

a user type determination device for determining a user type of the user by using words obtained based on the input,

the response determination device determining a commodity or a service to be presented to the user based on the determined user type of the user, the response determination device controlling information pertinent to the determined commodity or service so as to be included in the contents of the response.

By the information provision system according to the present invention, through a dialogue with a user utilizing an Internet shopping site, preference of the user is determined, and based on said determined preference, recommended commodities are determined. Therefore, the preference of the user is grasped in a specific and detailed manner, and commodities matching such preference are presented as recommended commodities. In addition, the commodities matching such preference are presented as the recommended commodities, thereby allowing various advantages in Internet shops, such as enhancing of sales per customer (including promotion of up-selling and cross-selling), boosting of profitability, and an increase in repeat users, to be expected.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a diagram illustrating an outline of an information provision system according to one embodiment of the present invention;

FIG. 2 is a diagram exemplifying a top screen of an Internet shopping site, according to the one embodiment of the present invention;

FIG. 3 is a diagram exemplifying a conversation screen on the Internet shopping site, according to the one embodiment of the present invention;

FIG. 4 is a diagram exemplifying a recommendation history display screen on the Internet shopping site, according to the one embodiment of the present invention;

FIG. 5 is a diagram exemplifying a conversation history display screen on the Internet shopping site, according to the one embodiment of the present invention;

FIG. 6 is a functional block diagram showing a user terminal in the information provision system according to the one embodiment of the present invention;

FIG. 7 is a functional block diagram showing an information provision server in the information provision system according to the one embodiment of the present invention;

FIG. 8 is a diagram for explaining in details functions of one part of the information provision server in the information provision system according to the one embodiment of the present invention;

FIG. 9 is a diagram showing an example of member information data in the information provision system according to the one embodiment of the present invention;

FIG. 10 is a diagram showing an example of commodity data in the information provision system according to the one embodiment of the present invention;

FIG. 11 is a diagram conceptually showing a method of determining recommended commodities in the information provision system according to the one embodiment of the present invention;

FIG. 12 is a diagram showing an example of a key word and picky-about item correspondence table in a first pattern of user type management in the information provision system according to the one embodiment of the present invention;

FIG. 13 is a diagram showing an example of a user type correspondence table in the first pattern of the user type management in the information provision system according to the one embodiment of the present invention;

FIG. 14 is a diagram conceptually showing a method of determining a user type in the first pattern of the user type management in the information provision system according to the one embodiment of the present invention;

FIG. 15 is a flowchart for explaining a processing procedure for determining a user type in the first pattern of the user type management in the information provision system according to the one embodiment of the present invention;

FIG. 16 is a diagram conceptually showing a method of determining recommended commodities in the first pattern of the user type management in the information provision system according to the one embodiment of the present invention;

FIG. 17 is a flowchart for explaining a processing procedure for determining recommended commodities in the first pattern of the user type management in the information provision system according to the one embodiment of the present invention;

FIG. 18 is a diagram showing an example of basic user types in a second pattern of the user type management in the information provision system according to the one embodiment of the present invention;

FIG. 19 is a diagram showing an example of applying a user to one of basic user types in the second pattern of the user type management in the information provision system according to the one embodiment of the present invention;

FIG. 20 is a diagram conceptually showing a method of determining a user type in the second pattern of the user type management in the information provision system according to the one embodiment of the present invention;

FIG. 21 is a diagram showing an example of basic user types in a third pattern of the user type management in the information provision system according to the one embodiment of the present invention;

FIG. 22 is a diagram showing an example of distribution of user types in the third pattern of the user type management in the information provision system according to the one embodiment of the present invention;

FIG. 23 is a diagram conceptually showing a method of determining a user type in the third pattern of the user type management in the information provision system according to the one embodiment of the present invention;

FIG. 24 is a diagram exemplifying a situation in which a user and a concierge have a conversation on a conversation screen in the information provision system according to the one embodiment of the present invention;

FIG. 25 is a diagram exemplifying a situation in which a user and the concierge have a conversation on a conversation screen in the information provision system according to the one embodiment of the present invention;

FIG. 26 is a diagram exemplifying a situation in which a user and the concierge have a conversation on a conversation screen in the information provision system according to the one embodiment of the present invention;

FIG. 27 is a diagram exemplifying a situation in which a user and the concierge have a conversation on a conversation screen in the information provision system according to the one embodiment of the present invention;

FIG. 28 is a diagram exemplifying a top screen of an Internet shopping site in the information provision system according to the one embodiment of the present invention;

FIG. 29 is a diagram exemplifying a situation in which a user and a concierge have a conversation on a conversation screen in the information provision system according to the one embodiment of the present invention;

FIG. 30 is a diagram exemplifying the situation in which the user and the concierge have the conversation on the conversation screen in the information provision system according to the one embodiment of the present invention;

FIG. 31 is a diagram exemplifying the situation in which the user and the concierge have the conversation on the conversation screen in the information provision system according to the one embodiment of the present invention;

FIG. 32 is a diagram exemplifying the situation in which the user and the concierge have the conversation on the conversation screen in the information provision system according to the one embodiment of the present invention;

FIG. 33 is a diagram exemplifying the situation in which the user and the concierge have the conversation on the conversation screen in the information provision system according to the one embodiment of the present invention;

FIG. 34 is a diagram exemplifying the situation in which the user and the concierge have the conversation on the conversation screen in the information provision system according to the one embodiment of the present invention;

FIG. 35 is a diagram exemplifying the situation in which the user and the concierge have the conversation on the conversation screen in the information provision system according to the one embodiment of the present invention;

FIG. 36 is a flowchart showing a typical pattern of a user serving processing procedure in the information provision system according to the one embodiment of the present invention;

FIG. 37 is a diagram for explaining how to determine recommendation timing at which a recommended commodity is displayed and how to conduct recommendation in the information provision system according to the one embodiment of the present invention; and

FIG. 38 is a diagram showing an example of a hardware configuration of a computer constituting the information provision server included in the information provision system according to the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT

Hereinafter, aninformation provision system1 according to one embodiment of the present invention will be described with reference to the accompanying drawings.

FIG. 1 is a diagram illustrating an outline of a system including theinformation provision system1 which responds to an input, thereby providing customer service and sales on an Internet shop.

Auser10 inputs a character string into an input area of a predetermined WEB page displayed by a WEB browser which is executed on a user terminal100 (refer to an upper portion inFIG. 1) (for example, by using a keyboard or a touch panel of the user terminal100). Here, the character string inputted by theuser10 is in the form of a text (input sentence) in general; includes, for example, questions and the like related to products of a company providing a WEB site having the above-mentioned WEB page; and is represented by a natural language. Hereinafter, for the sake of convenience, an input from a user is referred to as an “input sentence”. However, it is not required for the input from a user to be in the form of the text, and the input sentence means a variety of input character strings which include words and a part of a sentence.

In addition, here, the predetermined WEB page displayed by the WEB browser is, for example, a WEB page (site) of the so-called Internet shopping in which commodities are sold on the Internet. Theuser10 causes theuser terminal100 to display information pertinent to detailed information and a purchasing procedure of a commodity as a response from theinformation provision system1 according to the present embodiment and confirms the displayed information, thereby performing a conversation related to the commodity and the purchasing procedure.

It is to be noted that although in the present embodiment, the site of the Internet shopping on which the commodities are sold is described as an example, various sites on which convenience is provided for a user, such as sites of the Internet shopping on which services are provided and reservation or the like is made, are included as targets of the present invention.

The character string inputted by the user is provided via anetwork300 to theinformation provision system1 by the WEB browser of theuser terminal100. Here, thenetwork300 is, for example, a network including the Internet.

Upon receiving a character string from theuser terminal100, theinformation provision system1 determines a character string (a response sentence in general) and a commodity (group) as a response corresponding to the character string and transmits this response via thenetwork300 to theuser terminal100. As described above, when theuser10 has input a character string in the input area of the WEB page, the response (character string) and information or the like related to a commodity corresponding thereto are displayed in an output area of the WEB page. For example, when theuser10 has input a question related to a commodity in the input area of the WEB page, an answer to the question is displayed in an response display area of the WEB page, and in some case, information or the like related to the commodity is displayed in an commodity display area.

Theinformation provision system1 is configured to include aninformation provision server200. Theinformation provision server200 is provided with an information provision management DB (database)250 including member information data and commodity data. Theinformation provision server200 determines a character string having contents which are appropriate as a response corresponding to the input received from theuser terminal100 based on data stored in the informationprovision management DB250 and transmits response data to theuser terminal100 to cause the character string of said response to be displayed in the response display area (a response display area of a WEB page displayed by the WEB browser executed on the user terminal100).

In addition, in response to the input received from theuser terminal100, theinformation provision server200 determines information related to a recommended commodity by using this informationprovision management DB250 and transmits response data to theuser terminal100 to cause the information related to the commodity (group) to be displayed in the commodity display area (a commodity display area of a WEB page displayed by the WEB browser executed on the user terminal100).

Functions of theinformation provision server200 and contents of the informationprovision management DB250 will be described later in details.

Here, when theinformation provision system1 is provided as a system which can be used by unspecified users, as theuser10, general users who ask questions to companies and organizations (related to commodities, services, and the like thereof) which provide WEB sites are supposed, and theuser terminal100 is a PC (personal computer) or the like in general which can be connected to the Internet or the like and theuser10 has.

InFIG. 1, the configuration is arranged such that theuser10 inputs a character string in the input area of the WEB site by using a keyboard or the like, thereby providing theinformation provision system1 with a question and the like. However, any method or any route may be employed to input a character string to theinformation provision system1. For example, when theuser10 issues a question by voice, this voice is subjected to voice recognition processing to be converted to a character string, and said converted character string can be provided for theinformation provision system1. In addition, a printed question or the like is converted to a character string by using OCR (Optical Character Recognition), and said converted character string can also be provided for theinformation provision system1.

It is to be noted that theinformation provision server200 according to the present embodiment also has a function of a WEB server which is operable to perform data transmission and reception by a WEB browser executed on theuser terminal100 and a protocol such as HTTP and to cause said WEB browser to display a predetermined WEB page.

In addition, although theinformation provision server200 is shown as one computer here, theinformation provision server200 can also be configured as a plurality of computers, thereby allowing the same function to be shared and executed by the plurality of computers. In addition, the informationprovision management DB250 can also be configured such that the data stored therein is divided into various units to be shared by a plurality of sites and computers.

Further, in the present embodiment, a plurality ofinformation provision servers200 can be arranged, and each of theinformation provision servers200 is associated with one Internet shopping site or one site of a company providing commodities or the like. One Internet shopping site can also be subdivided to be realized by the plurality ofinformation provision servers200. Conversely, functions of a plurality of Internet shopping sites can also be realized by oneinformation provision server200.

FIG. 2 is a diagram exemplifying atop screen110 of an Internet shopping site. Theuser10 activates a WEB browser on theuser terminal100, inputs a URL of the Internet shopping site there (selects the URL thereof from bookmarks, directly inputs the URL, and so forth), and displays thetop screen110 shown inFIG. 2 on a display of theuser terminal100.

On thetop screen110, a topscreen display part111 is arranged, and in an upper portion therein, a title is displayed and on a left side, a list of categories is displayed. Theuser10 browses a variety of pieces of information displayed on this topscreen display part111 and clicks a linked text or image by using a mouse or the like, thereby causing information of a commodity to be displayed in a further detailed manner or conducting the procedure of purchasing a desired commodity. In addition, on a right side in the topscreen display part111, aconcierge image112 provided by theinformation provision system1 according to the present embodiment is shown. Theuser10 clicks thisconcierge image112 by using the mouse or the like to select theconcierge image112, whereby theuser10 can have a conversation (theinformation provision system1 responds to an input from the user10) with theinformation provision system1 with respect to the purchase of a commodity and the like.

In theinformation provision system1 according to the present embodiment, theconcierge image112 represents a concierge (a guide or an agent) who provides theuser10 with a variety of pieces of information (by using artificial intelligence or the like).

FIG. 3 is a diagram illustrating aconversation screen120 displayed by the WEB browser on theuser terminal100 after theuser10 has selected theconcierge image112 in the topscreen display part111 shown inFIG. 2. Although upon selecting theconcierge image112, a login window for inputting a user ID and a password is displayed, here, the login window is omitted.

On theconversation screen120 shown inFIG. 3, in an upper portion, aconversation display part121 is arranged, and on a right side in the upper portion, aconcierge image122 is displayed. Below theconversation display part121, aninput part124 in which a character string typed by theuser10 is inputted is arranged, and on a right side thereof, a “Speak”button125 which functions as a transmission button for transmitting the character string inputted by theuser10 to theinformation provision system1 is arranged.

Further, below theinput part124, acommodity display part126 in which a list of commodities recommended by theinformation provision system1 to theuser10 is displayed is arranged. In this example, in acommodity icon126A, an image of a commodity S1 is shown, and therebelow, an “Add to Cart” button with which a commodity is put into a cart (a shopping basket temporarily used for purchasing) is displayed. In addition, in acommodity icon126B, an image of a commodity S2 is shown, and therebelow, an “Add to Cart” button with which a commodity is put into a cart is displayed; in acommodity icon126C, an image of a commodity S3 is shown, and therebelow, an “Add to Cart” button with which a commodity is put into a cart is displayed; and in acommodity icon126D, an image of a commodity S4 is shown, and therebelow, an “Add to Cart” button with which a commodity is put into a cart is displayed.

Upon clicking the commodity icon such as thecommodity icon126A by the mouse or the like, detailed information of the corresponding commodity is displayed.

In the lowermost portion of theconversation screen120, abutton display part128 is arranged, and in thebutton display part128, arecommendation history button128A, aconversation history button128B, aHELP button128C, and aThanks button128D are displayed. Here, therecommendation history button128A is a button for displaying a recommendation history display screen (refer toFIG. 4); theconversation history button128B is a button for displaying a conversation history display screen (refer toFIG. 5); theHELP button128C is a button for displaying a HELP screen (not shown) related to theinformation provision system1; and theThanks button128D is a button clicked when theuser10 appreciates or evaluates a response from theinformation provision system1. A number of times at which theThanks button128D is clicked is used to determine a degree of satisfaction of theuser10.

FIG. 4 is a diagram illustrating a recommendationhistory display screen130 displayed by the WEB browser on theuser terminal100 when on theconversation screen120 shown inFIG. 3, therecommendation history button128A is selected by theuser10.

On the recommendationhistory display screen130 shown inFIG. 4, a recommendationhistory display part131 in which a list of commodities so far recommended as responses by theinformation provision system1 to theuser10 is displayed is arranged, and on a right side thereof, aconcierge image132 is displayed. It is to be noted that in this example, the most recent 100 commodities so far recommended are displayed, and other commodities other than those shown therein can be displayed by using a scroll bar displayed on a right side in the recommendationhistory display part131.

In the lowermost portion of the recommendationhistory display screen130, as with theconversation screen120, abutton display part133 is arranged, and in thebutton display part133, a “Speak with a concierge”button133A, aconversation history button133B, and a HELP button133C are displayed. Here, the “Speak with a concierge”button133A is a button to shift to a mode in which a conversation with the concierge is made; and theconversation history button133B and the HELP button133C are the same as theconversation history button128B and theHELP button128C of theconversation screen120 shown inFIG. 3, respectively.

FIG. 5 is a diagram illustrating a conversationhistory display screen140 displayed by the WEB browser on theuser terminal100 when on each of theconversation screen120 shown inFIG. 3 and the recommendationhistory display screen130 shown inFIG. 4, each of the conversation history buttons (128B and133B) is selected by theuser10.

On the conversationhistory display screen140 shown inFIG. 5, a conversationhistory display part141 in which contents of responses so far displayed to theuser10 as responses from theinformation provision system1 and contents inputted by the user10 (contents shown in theconversation display part121 of theconversation screen120 shown inFIG. 2) are displayed as the history is arranged, and on a right side thereof, aconcierge image142 is displayed. It is to be noted that in this example, all of the history of the conversation can be displayed, and contents not displayed in the conversationhistory display part141 can be browsed by using a scroll bar displayed on a right side in the conversationhistory display part141.

In the lowermost portion of the conversationhistory display screen140, as with theconversation screen120 and the recommendationhistory display screen130, abutton display part143 is arranged, and in thebutton display part143, a “Speak with a concierge”button143A, arecommendation history button143B, and aHELP button143C are displayed. The respective buttons are the same as those described with respect to theconversation screen120 and the recommendationhistory display screen130.

Next, with reference toFIG. 6, a configuration of theuser terminal100 will be described.FIG. 6 is a functional block diagram showing theuser terminal100.

Theuser terminal100 shown inFIG. 6 includes aninput control part101, aWEB browser102, and a network I/F (interface)part103.

When theuser10 has operated an input device such as a mouse, a keyboard, and a touch panel connected to theuser terminal100, theinput control part101 receives a signal generated by said operation as input information. TheWEB browser102 displays a specified WEB page and causes the input information received by theinput control part101 to be displayed in an input area of the WEB page. In addition, in response to the operation by the user10 (for example, clicking or the like of the “Speak”button125 functioning as the transmission button), an HTTP request is transmitted via the network I/F part103 to theinformation provision system1 functioning as a WEB server.

Further, upon receiving HTML data or the like including response data via the network I/F part103 from theinformation provision server200 or the like of theinformation provision system1, based on the data, a WEB page is displayed (a display of the WEB browser is updated).

The network I/F part103 is connected to thenetwork300 and controls data transmission and reception to and from theinformation provision server200 or the like of theinformation provision system1.

FIG. 7 is a functional block diagram showing theinformation provision server200 in theinformation provision system1.

Theinformation provision server200 shown inFIG. 7 includes aninput reception part201, aninput analysis part202, aresponse control part203, a commoditypurchase control part204, a recommendation history and conversationhistory management part205, a WEBaccess control part206, and a network I/F (interface)part207.

Theresponse control part203 further includes aninput evaluation part203A, a recommendation relateddata updating part203B, and aresponse determination part203C.

In addition, theinformation provision server200 is provided with an informationprovision management DB250. The informationprovision management DB250 includesdictionary data251,member information data261,commodity data262, usertype determination data263, preferredcommodity data264,recommendation history data271, purchasehistory data272,conversation history data273,scenario data281,FAQ answer collection282,HTML data291, andimage data292.

Theinput reception part201 receives a character string inputted by theuser10 onto theuser terminal100 via thenetwork300 and the network I/F part207, for example, according to a protocol such as HTTP.

Theinput analysis part202 extracts key words through natural language processing using the dictionary data251 (for example, key word extraction by a morphological analysis or the like is conducted).

Theinput evaluation part203A of theresponse control part203 stores a conversation constituted of inputs from theuser10 and responses by theinformation provision system1 and conducts conversion and supplementation of the extracted key words as needed.

The recommendation relateddata updating part203B of theresponse control part203 determines a user type of theuser10 and updates as needed user attribute data used when theresponse determination part203C determines recommended commodities for theuser10.

Theresponse determination part203C of theresponse control part203 determines recommended commodities and determines a response sentence and information provided for theuser10 in the form of responses from the concierge based on a responding policy.

When theuser10 purchases a commodity, the commoditypurchase control part204 controls a procedure of purchasing and guiding of a procedure of purchasing by theresponse determination part203C by using thescenario data281 or the like. In addition, the commoditypurchase control part204 stores the commodity purchased by theuser10 in thepurchase history data272.

The recommendation history and conversationhistory management part205 stores recommended commodities determined by theresponse determination part203C in therecommendation history data271 and stores contents of the conversation by theuser10 and the concierge in theconversation history data273. In addition, as shown inFIG. 4, in response to the operation by theuser10, the recommendation history and conversationhistory management part205 performs control so as to display recommended commodity information stored in therecommendation history data271 on the recommendationhistory display screen130 and as shown inFIG. 5, in response to the operation by theuser10, to display conversation information stored in theconversation history data273 on the conversationhistory display screen140.

When an HTTP request is transmitted from the WEB browser on theuser terminal100 in response to the operation by theuser10, the WEBaccess control part206 receives and analyzes this HTTP request, prepares data corresponding thereto, and sends a reply as an HTTP response to the WEB browser on theuser terminal100. The data sent as the HTTP response includes HTML data for, for example, displaying a recommended commodity, generated by theresponse determination part203C of theresponse control part203, and in addition thereto, HTML data generated for causing the recommendation history and conversationhistory management part205 to display the recommendationhistory display screen130 and the conversationhistory display screen140. In addition, the data sent as the HTTP response also includes theHTML data291 and theimage data292 in the informationprovision management DB250. TheHTML data291 is HTML data for displaying a WEB page and theimage data292 is image data and moving image data displayed in the WEB page.

The network I/F (interface)part207 is connected to thenetwork300 and controls data transmission and reception to and from theuser terminal100 or the like.

It is to be noted that although in the present specification, it is described that the variety of pieces of data including themember information data261 and the like are stored in the informationprovision management DB250, said variety of pieces of data can be stored as data having a variety of structures and formats including a table of a relational database and a flat file.

In addition, theinformation provision server200 can also be configured such that theinformation provision server200 itself does not have at least one part of the data, which is shown as being stored in the informationprovision management DB250 inFIG. 7 (for example, thecommodity data262, thescenario data281, and the like), and obtains these pieces of data externally (for example, from an external server connected via a network such as the Internet) by using predetermined commands.

FIG. 8 is a diagram for explaining the functions of one part of theinformation provision server200 in a further detailed manner. InFIG. 8, the respective function parts of theresponse control part203 shown inFIG. 7 (namely, theinput evaluation part203A, the recommendation relateddata updating part203B, and theresponse determination part203C) in theinformation provision server200 are shown.

Theinput evaluation part203A stores a conversation constituted of inputs from theuser10 and responses by theinformation provision system1 in theconversation history data273. Further, theinput evaluation part203A converts, or complements, the extracted key word to a word which is easily hit upon searching, interprets said key word, and converts, or complements, said key word to be in the form so as to allow theinformation provision system1 to easily handle said key word (for example, in the form in which theinformation provision system1 easily utilizes thescenario data281 and the like).

The recommendation relateddata updating part203B determines a user type of theuser10 based on the received utterance of theuser10 and data obtained from themember information data261. Contents of the utterance of theuser10 can be obtained from theconversation history data273 and can also be obtained via theinput reception part201 from theuser terminal100 at near-real-time timing. In addition, upon determining the user type, a key word and picky-about item correspondence table (refer toFIG. 12) and a user type correspondence table (refer toFIG. 13) which are stored in the usertype determination data263 are referenced.

In addition, with reference to a variety of pieces of data including picky-about item scores and basic user types stored in the usertype determination data263, thepreferred commodity data264, therecommendation history data271, thepurchase history data272, and the like, a user type can be determined. The determined user type of theuser10 is stored in themember information data261 so as to be associated with saiduser10.

In addition, the recommendation relateddata updating part203B stores and updates a variety of pieces of user attribute data as needed. For example, the recommendation relateddata updating part203B stores picky-about item scores related to eachuser10 as usertype determination data263. Further, the recommendation relateddata updating part203B stores data of user preferred commodities, user preference targets, purchase examining commodities, a commodity browse history, and the like related to eachuser10 as preferredcommodity data264. The user attribute data stored in this way, as described above, is used when theresponse determination part203C determines a recommended commodity for auser10.

Theresponse determination part203C obtains a user type of theuser10 from themember information data261, obtains information pertinent to purchased commodities related to a user whose user type is the same as the user type of theuser10 from thepurchase history data272, and based on these pieces of information, determines a recommended commodity to be presented to theuser10 from among commodities stored in thecommodity data262. The determined recommended commodity is stored in therecommendation history data271.

In addition, upon determining a recommended commodity, theresponse determination part203C can determine a commodity appropriate for theuser10 by utilizing theconversation history data273, thepreferred commodity data264, therecommendation history data271, and the like. In addition, based on purchased commodities, purchase examining commodities, browsed commodities, and the like related to not only the user whose user type is the same as the user type of theuser10 but also a user of a user type similar thereto, recommended commodities can be determined.

Further, with reference to thescenario data281 and the like, theresponse determination part203C determines a responding policy for theuser10 and determines contents of a response to theuser10 based on data such as thescenario data281 and the FAQ answer collection. In addition, with reference to a variety of pieces of data such as theconversation history data273, themember information data261, therecommendation history data271, thepurchase history data272, and thecommodity data262, based on these pieces of data, theresponse determination part203C can determine the contents of a response. In thescenario data281, response patterns corresponding to input patterns of theuser10 are defined, and based on these response patterns, the responding policy and the contents of a response are determined. Thescenario data281 includes knowledge data, and said knowledge data can be configured to be updated by learning through the conversation with theuser10.

By the above-describedresponse determination part203C, theinformation provision system1 not merely receives an input from theuser10 and returns a passive response in accordance with the contents thereof but also can output, to theuser terminal100, as a response an active question which draws out additional key words and information from theuser10 in order to determine recommended commodities and the like in a more specific manner with high precision.

In addition, a scenario is configured such that users are purposefully guided to surplus inventory commodities in theinformation provision system1 and commodities whose profitability ratios are high, thereby allowing said commodities to be presented as recommended commodities in a conversation with theuser10.

It is to be noted that although inFIG. 8, accesses of the respective function parts which are theinput evaluation part203A and the others with the variety of pieces of data are indicated by arrows, these merely indicate accesses related to representative processes of the respective function parts, and the present invention is not limited thereto.

FIG. 9 is a diagram showing an example of themember information data261 included in the informationprovision management DB250 utilized by theinformation provision server200 or the like in theinformation provision system1.

As shown inFIG. 9, in themember information data261, for example, a user ID, a password, a user type, a name, sex, an occupation, age, the date of birth, an habitation area, marriage, hobbies, interesting categories, uninteresting categories, events of personal interest, a purchase history, a purchase price, a number of times of visiting, a degree of satisfaction, and the like are stored.

Among these items, the user ID, the user type, the purchase history, the purchase price, the number of times of visiting, and the degree of satisfaction are system setting items which are automatically set and updated by theinformation provision system1, and the other items are user registration items which are registered when theuser10 makes membership registration.

The user ID is an ID which is automatically numbered by theinformation provision system1 when theuser10 initially makes the membership registration. The password is inputted by theuser10 upon logging-in and is set by theuser10 when theuser10 initially makes the membership registration.

Theinformation provision system1 classifies respective users as user types based on contents of conversations of theusers10, member information such as sex and age, and the like. The user type is data (for example, “A”, “C”, and the like) indicating a user type determined for each of the respective users as belonging thereto. The determination of the above-mentioned user types will be described later in details.

The name is a name of himself or herself which is set by theuser10 and may be a nickname or the like. The sex, the occupation, the age, the date of birth, the habitation area, the marriage, the hobbies, the interesting categories, the uninteresting categories, and the events of personal interest are selected from choices shown in theinformation provision system1 to be set when theuser10 initially makes the membership registration or the like.

In the purchase history, based on thepurchase history data272 in which the commodities so far purchased by theuser10 are recorded, for example, the commodities purchased last time, categories of often purchased commodities, and the like are set. In the purchase price, based on the above-mentionedpurchase history data272, for example, an average price of purchased commodities, a total amount of prices of purchased commodities, and the like are set.

In the number of times of visiting, based on theconversation history data273 in which a history of the conversation which theuser10 has so far had with theinformation provision system1 is recorded, for example, data for distinguishing between a frequency of five to nine times and a frequency of ten times or more is set. In the degree of satisfaction, for example, a number of times at which theThanks button128D displayed on theconversation screen120 shown inFIG. 3 is clicked is counted, and the counted number is set.

FIG. 10 is a diagram showing an example of thecommodity data262 included in the informationprovision management DB250 utilized by theinformation provision server200 or the like in theinformation provision system1.

As shown inFIG. 10, in thecommodity data262, for example, a commodity ID, a commodity name, a maker name, a commodity category, detailed information, recommendation information, discount information, customer information, new model information, economy-size model information, and the like are stored.

In the commodity category, categories classified by a plurality of criteria can be included. For example, the commodity category can be configured so as to include commodity categories by basic classification based on commodity classification and by-purpose categories by classification based on purposes of using commodities.

The detailed information is, for example, information pertinent to content quantities, ingredients, explanation about commodities, and the like. The recommendation information includes words of recommending said products, seasonal recommendation messages, and the like, which can also be prepared by each maker and each Internet shop. The discount information is information pertinent to discount and campaigns, which an Internet shop prepares and adds. The discount information may be associated with a plurality of commodity groups, a predetermined category, and other classification, instead of each commodity.

In the customer information, for example, customers' comments and purchasers' use feeling are stored for each commodity. In the new model information, for example, new models of commodities and information pertinent to the new models are stored. In the economy-size model, for example, commodity IDs of economy-size commodity models, which are commodities with larger economy sizes, are stored.

Next,FIG. 11 is a diagram conceptually showing how in theinformation provision system1 according to the present embodiment, recommended commodities to be presented to theuser10 are determined. The presentation of the recommended commodities to theuser10 by theinformation provision system1 is conducted, for example, in a case where when “Mr. (or Ms.) XX” who is theuser10 inputs (utters) “I want a stylish table.” in theconversation display part121 of theconversation screen120 shown inFIG. 3, a “concierge” who is an agent of theinformation provision system1 responds, saying “How about ones like these?” and concurrently displays the recommended commodities in thecommodity display part126.

InFIG. 11, when the recommended commodities are determined, a category of commodities is first selected and thereafter, the recommended commodities are selected from among commodities purchased by a user whose user type is the same as the user type of theuser10 who has uttered based on predetermined criteria.

For example, a first recommended commodity is a commodity belonging to a category which theuser10 is currently browsing in a WEB page and is determined by identifying a commodity which ranks first among commodities purchased by the other user whose user type is the same as the user type of thisuser10. It is to be noted that as in theconversation display part121 of theconversation screen120 shown inFIG. 3, when a specific commodity category (in this example, the “table”) is cited, based on the cited content of the utterance, commodities are searched from the commodity category (table). In addition, in this example, the ranking shows a ranking within the most recent one week.

As described above, in theinformation provision system1 according to the present embodiment, based on the purchase information related to other user whose user type is the same as the user type of the user10 (for example, information pertinent to what commodities theuser10 is interested in), recommended commodities can be determined. As the purchase information, information pertinent to commodities actually purchased by other user (including hot-selling ranking or the like) is included. In addition, information pertinent to behavior of other user before purchasing such as information pertinent to commodities browsed in a WEB page and information pertinent to commodities which are examined for the purchase in the conversation with theinformation provision system1 is also included as the purchase information.

Next, with reference toFIG. 12 toFIG. 15, a first pattern of user type management in theinformation provision system1 according to the present embodiment will be described.

FIG. 12 is a correspondence table (key word and picky-about item correspondence table) in which key words and picky-about items are associated with each other. Utterance of theuser10 inputted to theinformation provision system1 is subjected to natural language processing using thedictionary data251 or the like, and key words related to determination of a user type as shown in the key word and picky-about item correspondence table inFIG. 12 are extracted. In the key word and picky-about item correspondence table inFIG. 12, key words are considered to show preference of a user, respective key words and kinds of pickiness related thereto are set. It is to be noted that in the present specification, the key words embrace concept such as phrases and sentences including a plurality of words.

For example, when theuser10 utters key words, “marine-taste”, “adorable”, and “it is cool” in a conversation with the concierge, theuser10 is evaluated as being picky about a design, and said user is associated with a picky-about item of a “design” (NO. 1).

In addition, when theuser10 utters key words, “ecological”, “quick”, and “excellent effect” in a conversation with the concierge, theuser10 is evaluated as being picky about functionality, effect, and efficacy, and said user is associated with a picky-about item of “functionality, effect, and efficacy” (NO. 2).

FIG. 13 is a user type correspondence table showing association of user types (here, A to X) based on picky-about item NOs (picky-about items NO. 1 to NO. 10 inFIG. 12 shown above), sex, and age.

For example, when auser10 whose picky-about item is evaluated to be a picky-about item NO. 1 (design) with reference to the key word and picky-about item correspondence table inFIG. 12 is a male in his 10's, a user type of theuser10 is determined to be “G”. A user of the user type “G” is a user who has, as attributes, great pickiness about a design, feeling, and taste (in setting in this example).

In addition, when auser10 whose picky-about item is evaluated to be a picky-about item NO. 2 (functionality, effect, and efficacy) with reference to the key word and picky-about item correspondence table inFIG. 12 is a female in her 30's, a user type of theuser10 is determined to be “E”. A user of the user type “E” is a user who has, as attributes, great pickiness about functionality, effect, efficacy, usability, and quality (in setting in this example).

FIG. 14 is a diagram conceptually showing how the user type is determined with respect to eachuser10 in theinformation provision system1 according to the present embodiment. As shown inFIG. 14, based on data inputted by thatuser10 to the user terminal100 (for example, utterance of theuser10 included in theconversation history data273, which includes utterance before purchasing a commodity and utterance after purchasing the commodity) and member information (for example, data pertinent to sex and age included in the member information data261), a user type of theuser10 is determined (refer to the user type correspondence table shown inFIG. 13).

In addition, as shown inFIG. 14, based on the user type determined as described above, a scenario with respect to the utterance of theuser10 is searched, and in accordance with the scenario, a response is generated.

In addition, since the user type of auser10 shows attributes of a user, a plurality of user types can also be set for eachuser10. In addition, since the utterance of theuser10 is accumulated each time theuser10 accesses theinformation provision system1, such contents of the utterance are extracted at predetermined timing and a user type or user types can also be thereby re-examined.

FIG. 15 is a flowchart for explaining a processing procedure for determining a user type from the utterance of theuser10 based on the key word and picky-about item correspondence table shown inFIG. 12 and the user type correspondence table shown inFIG. 13. The processing shown inFIG. 15 is called up, for example, each time theuser10 utters.

First, at step S11, theinformation provision server200 of theinformation provision system1 receives an input (utterance) of theuser10 from theuser terminal100 and through the natural language processing or the like using thedictionary data251 or the like, extracts key words from the input. Next, at step S12, it is determined whether or not key words to be extracted are present. When the key words are not present (NO at step S12), the processing is finished.

When the key words are present (YES at step S12), at step S13, it is determined whether or not the extracted key words correspond to any of the picky-about items (that is, it is determined whether or not any picky-about item corresponding to any of key words is present with reference to the key word and picky-about item correspondence table inFIG. 12). When the extracted key words do not correspond to any of the picky-about items (NO at step S13), the processing returns to step S11 and processing for the next key words is conducted.

When the extracted key words correspond to any of the picky-about items (YES at step S13), at step S14, frequencies at which key words correspond any of the picky-about items are totalized for each user. The totalized frequencies of the picky-about items are stored, for example, as the usertype determination data263.

Thereafter, at step S15, it is determined whether or not the totalized frequencies of the picky-about items satisfy predetermined criteria. When the predetermined criteria are not satisfied (NO at step S15), the processing returns to step S1land processing for the next key words is conducted. When the predetermined criteria are satisfied (YES at step S15), at step S16, with reference to themember information data261, age and sex of the correspondinguser10 are obtained.

Next, at step S17, with reference to the user type correspondence table (the user type correspondence table shown inFIG. 13), based on a combination of the picky-about items determined as satisfying the predetermined criteria and the age and sex of the user obtained at step S16, a user type is obtained and determined as the user type of theuser10.

Thereafter, at step S18, the determined user type is set as the corresponding user type of theuser10 in the member information data261 (updated in a case where the determined user type has already been set).

It is to be noted that although in this example, when theuser10 utters the key words corresponding to the picky-about items several times, it is determined that the predetermined criteria are satisfied, and these picky-about items are set as the picky-about items of theuser10, by employing a method other than this, based on the utterance of theuser10, picky-about items of theuser10 may be determined.

Next, with reference toFIG. 16 andFIG. 17, a method of determining recommended commodities in theinformation provision system1 according to the present embodiment will be described.

FIG. 16 is a diagram conceptually showing how the recommended commodities are determined. As shown inFIG. 16, in this example, based on criteria related to commodity categories and criteria related to commodities purchased by other user whose user type is the same as the user type of theuser10, or only based on the criteria related to the commodities purchased by the other user whose user type is the same as the user type of theuser10, recommended commodities are determined and presented to theuser10. It is to be noted that in this example, the commodity category is the commodity category which theuser10 is displaying on a WEB page.

In addition, for the determination of the above-described recommended commodities, other criteria related to commodity categories can also be used and other criteria related to commodities purchased by other user whose user type is the same as the user type of theuser10 can also be used. Further, for the determination of the recommended commodities, criteria other than the criteria related to the commodity categories and other than the criteria related to the commodities purchased by the other user whose user type is the same as the user type of theuser10 can also be used.

Further, in this example, the recommended commodities are determined based on the purchase information of other user whose user type is the same as the user type of theuser10. However, based on purchase information related to other user whose user type is similar to the user type of theuser10 and purchase information related to other user who is considered, from a predetermined viewpoint, to belong to the same classification as classification to which theuser10 belongs, recommended commodities can be determined. For example, as in the user type correspondence table shown inFIG. 13, user types A, B, C, D, E, and F can be considered to belong to the same classification in that users of these user types have common attributes. In addition, user types G, H, and I can be considered to belong to the same classification in that the sex thereof is the same as one another and users of these user types have common attributes.

Next, with reference to a flowchart inFIG. 17, based on a user type of theuser10 determined in the processing procedure shown in the flowchart inFIG. 15, a processing procedure for determining recommended commodity for theuser10 will be described. It is to be noted that in this example, a pattern in which based on a commodity category which theuser10 is browsing and a user type, recommended commodity are determined will be described.

First, at step S21, utterance of theuser10 is received, and from the utterance, recommendation timing is determined. In this processing, for example, from the utterance of theuser10, key words are extracted through natural language processing or the like using thedictionary data251 or the like, and from the extracted key words and thescenario data281 or the like, the recommendation timing at which recommended commodities are presented to theuser10 is determined. Next, at step S22, it is determined whether or not the present time is recommendation timing, and when the present time is not the recommendation timing (NO at step S22), the processing returns to step S21 and utterance of theuser10 is received.

When the present time is the recommendation timing (YES at step S22), at step S23, a commodity category in a WEB page which theuser10 is currently browsing is obtained. Next, at step S24, from themember information data261, a user type is obtained. Here, since theuser10 has logged in by using a user ID upon browsing the WEB page for Internet shopping, thecurrent user10 can be identified therefrom and the user type corresponding thereto can be obtained from themember information data261.

Next, at step S25, with reference to thepurchase history data272, ranking of commodities purchased by the user of the same user type, with respect to the same (or different) category as the obtained commodity category, is obtained. Here, when thepurchase history data272 has only the purchase history recorded therein, all of the related purchase history is obtained, and from that history, ranking data is generated.

Thereafter, at step S26, a commodity in predetermined ranking (for example, in the commodity category currently being browsed, a commodity which ranks first among commodities purchased by a user who user type is the same as the user type of the user10) is determined as a recommended commodity, and HTML data for presenting the determined recommended commodity to theuser10 is generated.

Next, with reference toFIG. 18 toFIG. 20, a second pattern in the user type management in theinformation provision system1 according to the present embodiment will be described.

In the second pattern, as shown inFIG. 18, three basic user types (TYPE-1, TYPE-2, and TYPE-3) are prepared, and with respect to each of the basic user types, degrees of picky-about items are defined from three viewpoints.

For example, with respect to the TYPE-1, degrees of picky-about items are defined such that a degree of being picky about a product is low (a score: 20 to 40), a degree of being picky about a price is high (a score: 75 to 95), and a degree of accepting recommendation (in other words, a user of the TYPE-1 is in need of the presentation of recommended commodities provided from theinformation provision system1 and has a stance in which the user accepts said presentation) is high (a score: 75 to 95).

In addition, with respect to the TYPE-2, degrees of picky-about items are defined such that a degree of being picky about a product is very high (a score: 80 to 100), a degree of being picky about a price is very high (a score: 80 to 100), and a degree of accepting recommendation is extremely high (a score: 0 to 10). Further, with respect to the TYPE-3, degrees of picky-about items are defined such that a degree of being picky about a product is slightly low (a score: 30 to 50), a degree of being picky about a price is medium (a score: 40 to 60), and a degree of accepting recommendation is slightly high (a score: 50 to 60).

On a left side ofFIG. 19, with respect to an individual user type of theuser10, scores of picky-about items from the three viewpoints (that is, the pickiness about a product, the pickiness about a price, and the pickiness about recommendation) are shown. These scores of the picky-about items are adjusted by performing addition and subtraction based on key words extracted from utterance of theuser10. Each of these scores takes a value of 0 to 100. For example, each time theuser10 utters key words indicating that a degree of being picky about a product is low, subtraction from the score related to the degree of being picky about a product as the picky-about item is performed, and in this example, a value of the score is adjusted to be 28. Similarly, the score related to a degree of being picky about a price is adjusted to be 80 and the score related to a degree of accepting recommendation is adjusted to be 80, respectively.

On a right side ofFIG. 19, with respect to the basic user type “TYPE-1” shown inFIG. 18, under which the individual user type of theuser10 shown on the left side ofFIG. 19 falls and which is determined as the user type “TYPE-1” among the basic user types, ranges of scores thereof are shown. Specifically, with respect to the individual user type shown on the left side ofFIG. 19, the score related to the degree of being picky about a product is 28 and falls in a range of 20 to 40 of the score related to the degree of being picky about a product with respect to the basic user type “TYPE-1” shown on the right side thereof. In addition, with respect to the individual user type shown on the left side ofFIG. 19, the score related to the degree of being picky about a price is 80 and falls in a range of 75 to 95 of the score related to the degree of being picky about a price with respect to the basic user type “TYPE-1” shown on the right side thereof. Similarly, with respect to the individual user type shown on the left side ofFIG. 19, the score related to the degree of accepting recommendation is 80 and falls in a range of 75 to 95 of the score related to the degree of accepting recommendation with respect to the basic user type “TYPE-1” shown on the right side thereof. Of course, since each time theuser10 utters, the scores related to the picky-about items of the individual user type change, a corresponding basic user type is likely to change over time.

FIG. 20 is a diagram conceptually showing how each of the user types related to theuser10 is determined in the above-described second pattern. As shown inFIG. 20, based on data inputted by theuser10 onto the user terminal100 (for example, utterance by theuser10 included in theconversation history data273, including utterance before purchasing a commodity and utterance after purchasing the commodity), the member information, and the like, scores of the respective picky-about items of the user type of theuser10 are adjusted.

In addition, also based on a result of an initial diagnostic test conducted for theuser1, the purchase history stored in thepurchase history data272, behavior related to the purchase, and the like, the scores can be adjusted. In addition, as initial values of a user type, scores of a predetermined basic user type can also be used.

For a predetermined period of time, the scores are adjusted as described above, and based on the adjusted scores, a basic user type which falls under any one among the user types is determined and is determined as the user type of theuser10.

In addition, as shown inFIG. 20, based on the basic user type determined as described above, a scenario with respect to the utterance by theuser10 is searched, and in accordance with the scenario, a response is generated.

In addition, since each time theuser10 accesses theinformation provision system1, theuser10 utters and the utterance is accumulated, such contents of the utterance are extracted at predetermined timing, and the user type can also be thereby re-examined.

It is to be noted that although in this example, based on the member information, the user type is determined, the user type can also be determined without using the member information.

Next, with reference toFIG. 21 toFIG. 23, a third pattern of the user type management in theinformation provision system1 according to the present embodiment will be described.

As shown inFIG. 21, in the third pattern, in accordance with degrees of picky-about items from two viewpoints, nine user types are set. One of the picky-about items is a product and the other thereof is a price. Each of the picky-about items is evaluated at four levels: “having pickiness”, “having moderate pickiness”, “having little or no pickiness”, and “having no pickiness”. In accordance with these combinations, nine user types are defined, and further, these plurality of user types are classified into several groups.

For example, as shown inFIG. 21, a user type having pickiness about a product and having little or no pickiness about a price (type A), a user type having pickiness about a product and also having pickiness about a price (type C), and a user type having pickiness about a price and having little or no pickiness about a product (type I) are classified as a “picky group”.

In addition, a user type having pickiness about a product and having moderate pickiness about a price (type B) and a user type having pickiness about a price and having moderate pickiness about a product (type F) are classified as a “partially picky group”.

In addition, a user type having no pickiness about a price and having moderate pickiness about a product (type D) and a user type having no pickiness about a product and having moderate pickiness about a price (type H) are classified as a “slightly picky group”.

In addition, a user type having moderate pickiness about a product and having moderate pickiness about a price (type E) is classified as a “normally picky group”. A user type having no pickiness about a product and having no pickiness about a price (type G) is classified as a “non-picky group”.

FIG. 22 is a graph showing distribution of the user types described above. Here, scores related to the basic user type E (normally picky group) are set as initial values upon determining a user type of theuser10.

FIG. 23 is a diagram conceptually showing how a user type with respect to eachuser10 is determined in the above-described third pattern. As shown inFIG. 23, based on data inputted by theuser10 onto the user terminal100 (for example, utterance by theuser10 included in theconversation history data273, including utterance before purchasing a commodity and utterance after purchasing the commodity) and the like, picky-about items of a user type of the user10 (degrees of picky-about items) are determined.

In addition, in this example, as the initial values of a user type, those of the user type E are used as mentioned above.

For a predetermined period of time, determination of degrees of picky-about items is conducted, and a user type corresponding to the determined degrees of the picky-about item is determined and is determined as a user type of theuser10.

In addition, as shown inFIG. 23, based on the user type determined as mentioned above, a scenario with respect to the utterance by theuser10 is searched, and in accordance with the scenario, a response is generated.

In addition, since the utterance of theuser10 is accumulated each time theuser10 accesses theinformation provision system1, such contents of the utterance are extracted at predetermined timing and a user type can also be thereby re-examined.

Next, with reference toFIG. 24, a situation in which in the commodity search by theuser10, “Mr. (or Ms.) XX” who is theuser10 and a “concierge who is the agent of theinformation provision system1 have a conversation will be specifically exemplified.FIG. 24 is a diagram showing an input by theuser10 and contents of a response from theinformation provision system1 displayed in theconversation display part121 of theconversation screen120 shown inFIG. 3. Here, for the sake of convenience, as theconversation display part121, aconversation display part601 which is vertically long is shown, and contents sequentially displayed by operating a scroll bar are concurrently displayed (the same is applied to a conversation display part shown hereinafter).

It is to be noted that in this example, the user type management in the above-described first pattern is used.

In the conversation displayed in theconversation display part601 shown inFIG. 24, first, in response to logging-in by theuser10, theinformation provision system1 makes aresponse602 such as “Welcome!”. Thereafter, theuser10 makes aninput603 saying “I am looking for an easy-to-carry headset.”

Here, theinformation provision system1 extracts words “easy-to-carry” from the input by theuser10 and converts key words such as “light-weight” and “compact” which are easily hit upon searching. In addition, as key words, a “headset” and a “wireless headset” are also extracted.

In addition, at this time, theinformation provision system1 sets a picky-about item of thisuser10 to be “usability” (NO. 6 shown inFIG. 12) based on the key word “compact” and determines a user type=“A” (theuser10 is a male in his 10's) from this picky-about item with reference to the user type correspondence table shown inFIG. 13.

Next, theinformation provision system1 displays a prompt to select a maker from a plurality of candidates (response604) and displays a list of makers of a headset in the commodity display part126 (thecommodity display part126 of theconversation screen120 shown inFIG. 3) (display processing605). At this time, theinformation provision system1 references thecommodity data262 or the like for searching and displays the list of makers of the headset, obtained as a result of searching.

Thereafter, when theuser10 selects amaker1 from the list of makers displayed (selection processing606), theinformation provision system1 references thepurchase history data272 or the like, searches headsets manufactured by themaker1, which are purchased by other user belonging to the user type=“A” and satisfy conditions of the key words “light-weight” and “compact”, and displays a list of the headsets obtained as a result of searching in the commodity display part126 (display processing607).

It is to be noted that when theinformation provision system1 displays the list of makers in thedisplay processing605, theinformation provision system1 may conduct narrowing-down based on a user type and conditions of key words. In addition, theinformation provision system1 can also make an inquiry as to color, a price range, and the like to theuser10 and conduct the narrowing-down of commodities based on these pieces of information. Further, theinformation provision system1 can also present (or sort) commodities whose models are the latest ones and whose prices are the lowest ones in accordance with preference of theuser10. The preference of theuser10 can be grasped, for example, based on a user type and by referencing thepreferred commodity data264 or the like.

Next, theinformation provision system1 prompts theuser10 to browse commodities in a detailed manner, in which theuser10 is interested (response608). In response thereto, theuser10 selects acommodity1 from among the displayed list of commodities (selection processing609).

In response to the selection processing by theuser10, theinformation provision system1 displays explanation regarding the selectedcommodity1, displays information pertinent to a new version of said model, and prompts theuser10 to browse a detailed page (response610). The explanation of thecommodity1 and the information pertinent to the new version of the model can be obtained, for example, with reference to recommendation information, new model information, and the like in thecommodity data262.

Thereafter, theinformation provision system1 controls thecommodity display part126 to display a WEB page of a maker, in which detailed explanation of thecommodity1 is given (display processing611). In addition, when theuser10 performs an operation to display the information regarding the new model, theinformation provision system1 sets a picky-about item of thisuser10 to be “hot-selling” (NO. 7 inFIG. 12), determines a user type=“S” from the picky-about item with reference to the user type correspondence table inFIG. 13, and updates themember information data261 by replacing the user type=“A” determined as above with the user type=“S” as a user type of the user10 (or setting the user type=“S” together with the user type=“A”). As described above, in accordance with not only the input by theuser10 but also various behavior including selection operation such as clicking operation by theuser10, theinformation provision system1 can determines a user type.

As described above, theinformation provision system1 extracts the key words utilized for determining a user type from the input by theuser10 and extracts the key words for narrowing down the commodities, displays the list of commodities satisfying conditions based on the user type and the key words, and thereby can provide the information pertinent to commodities which are appropriate for theuser10.

Next, with reference toFIG. 25, a situation in the scene of recommendation and collection of evaluation related to a purchased commodity in which “Mr. (or Ms.) YY” who is auser10 and a “concierge” who is an agent of theinformation provision system1 have a conversation will be more specifically exemplified.

In the conversation shown in aconversation display part621 inFIG. 25, first, in response to logging-in by theuser10, theinformation provision system1 greets theuser10, saying “Welcome!” and the like and subsequently, makes aresponse622, providing information pertinent to a campaign and asking evaluation as to the purchased commodity. Such contents of the response by theinformation provision system1 can be adjusted in accordance with a total amount of purchase prices and a number of times of visiting of the logging-inuser10. It is to be noted that the total amount of purchase prices and the number of times of visiting thereof by theuser10 are stored in themember information data261.

In response to thisresponse622, theuser10 inputs feedback as to the purchased commodity (input623), saying “it is good because it is very light-weight, compact, and convenient in carrying”. Here, theinformation provision system1 extracts key words “light-weight” and “compact” from this input, sets a picky-about item of theuser10 to be “usability” (NO. 6 inFIG. 12) based on the key word “compact”, and determines a user type=“D” from the picky-about item with reference to the user type correspondence table inFIG. 13 (theuser10 is defined as a female in her 20's).

Next,information provision system1 proposes introduction of recommended commodities to the user10 (response624), and theuser10 accepts this (input625). In response thereto, theinformation provision system1 recommends five commodities, makes a response that a wireless headset manufactured by amaker1 is particularly recommended among the recommended commodities (response626), and displays the five commodities in the commodity display part126 (display processing627). These recommended commodities are selected based on a user type of theuser10, an average purchase price, a number of times of visiting, hot-selling ranking, user attributes (in20's and female), and the like.

Further, here, when theinformation provision system1 makes an inquiry as to whether theuser10 browses details of the particularly recommended commodity (the wireless headset manufactured by the maker1) (response628), theuser10 expresses her intention to browse said details (input629). In response thereto, theinformation provision system1 controls thecommodity display part126 to display a WEB page of the maker in which detailed explanation of the recommended commodity is given (display processing630).

Next, with reference toFIG. 26, a situation in the scene of making a request by a customer in which “Mr. (or Ms.) XX” who is auser10 and a “concierge” who is an agent of theinformation provision system1 have a conversation will be more specifically exemplified.

In the conversation shown in aconversation display part641 inFIG. 26, first, theuser10 makes an inquiry as to whether “Noshi”, Japanese gift wrapping paper, can be attached to a commodity (input642). In response thereto, theinformation provision system1 displays a corresponding response sentence previously prepared based on contents of the input (response643) and displays, as a link text, a URL of a WEB page in which the “Noshi” is explained in details (link text644).

Thereafter, theinformation provision system1 prompts theuser10 to click a Thanks button when the problem is solved (response645), and in response to this, theuser10 clicks the Thanks button (selection processing646). A number of times at which the Thanks button is clicked is counted and is stored as a degree of satisfaction in themember information data261.

Next, with reference toFIG. 27, a situation in the scene of addressing a complaint in which “Mr. (or Ms.) XX” who is auser10 and a “concierge” who is an agent of theinformation provision system1 have a conversation will be more specifically exemplified.

In the conversation shown in aconversation display part661 inFIG. 27, first, theuser10 conveys a message regarding inadequacy of a commodity and makes an inquiry as to whether the commodity can be returned (input662). In response thereto, theinformation provision system1 displays a corresponding response sentence previously prepared based on contents of the input (response663), displays a screen of a telephone and a live chat in thecommodity display part126, and starts addressing by an operator (addressing processing664).

Next, with reference toFIG. 28 toFIG. 35, a situation, in the scene where theuser10 undergoes the commodity searching and purchase addressing by theinformation provision system1 and purchases commodities, in which “Mr. (or Ms.) ZZ” who is auser10 and a “concierge” who is an agent of theinformation provision system1 have a conversation will be more specifically exemplified.

FIG. 28 is a diagram illustrating atop screen701 of an Internet shopping site (AA on-line shop). Theuser10 activates a WEB browser on theuser terminal100, inputs a URL of the AA on-line shop (for example, selects the URL from bookmarks or the like or directly inputs the URL), and displays thetop screen701 inFIG. 28 on a display of theuser terminal100.

On thetop screen701, a topscreen display part702 is arranged, and in an upper portion thereof, a title is displayed and on a left side thereof, a list of categories is displayed. Theuser10 browses a variety of pieces of information displayed in this topscreen display part702 and clicks linked texts and images by using a mouse or the like, thereby causing information pertinent to commodities to be displayed in a further detailed manner and conducting a procedure of purchasing desired commodities.

On a right side of the topscreen display part702, a log-inpart703 for connecting to theinformation provision system1 according to the present embodiment is displayed, and theuser10 inputs a user ID and a password here and clicks a log-in button by using a mouse or the like. By clicking this log-in button, theuser10 can have a conversation with theinformation provision system1 as to purchasing of commodities and the like (in response to the input by theuser10, theinformation provision system1 responds).

Upon logging-in, shifting to a conversation screen is conducted, and this conversation screen is the same as theconversation screen120 shown inFIG. 3. Each ofFIG. 29 toFIG. 35 is a diagram showing contents of inputs by theuser10 and contents of responses by theinformation provision system1, displayed in a display part corresponding to theconversation display part121 of theconversation screen120 shown inFIG. 3. Here, for the sake of convenience, as the display part corresponding to theconversation display part121, a conversation display part which is vertically long is shown, and contents sequentially displayed by operating a scroll bar are concurrently displayed. In other words, conversation display parts from aconversation display part711 shown inFIG. 29 to aconversation display part771 shown inFIG. 35 constitute one connected display part.

It is to be noted that in this example, the user type management in the above-described second pattern is used.

In a conversation shown in theconversation display part711 inFIG. 29, first, theuser10 logs in the site of the AA on-line shop, and in response thereto, theinformation provision system1 greets theuser10, saying “Mr. (or Ms.) ZZ, Long time no see. Have you been well?” or the like and makes aresponse712 including announcement of a campaign and the like. Based on a nickname, campaign information, a previous browsing period of time, and the like, the above-mentioned greeting is generated based on a scenario.

Here, theuser10 makes aninput713 saying “I am thinking about the purchase of the commodity recommended before”. Here, based on the input by theuser10, theinformation provision system1 makes aresponse714 asking whether or not the commodity (commodity AA01) recommended before is put into a cart. The commodity recommended to theuser10 before is stored as a purchase examining commodity, for example, in therecommendation history data271 and thepreferred commodity data264. In addition, at this time, theinformation provision system1 deems that with respect to theuser10, a preference degree as to said commodity is increased and performs addition to a preference degree score of said commodity. The above-mentioned preference degree score can be recorded in thepreferred commodity data264.

Theinformation provision system1 displays information pertinent to the commodity AA01 together with the above-mentionedresponse714 in a commodity display part corresponding to thecommodity display part126 of theconversation screen120 shown inFIG. 3 (display processing715).

Here, theinformation provision system1 asks whether or not the commodity AA01 is put into the cart (response716), and in response thereto, theuser10 expresses her intention to put the commodity AA01 into the cart (input717).

At this time, theinformation provision system1 deems that theuser10 purchases the recommended commodity without having further pickiness and performs addition by a predetermined number to a score of the picky-about item “a degree of accepting recommendation” of theuser10. When due to a change in the above-mentioned score, theuser10 comes to belong to other user type, themember information data261 is updated so as to associate theuser10 with said changed user type.

In response to theinput717 by theuser10, theinformation provision system1 makes aresponse718 saying “Very glossy eyelashes are lovely, aren't they?” This is cited from, for example, recommendation information related to said commodity AA01 stored in thecommodity data262.

Here, theinformation provision system1 controls theuser terminal100 to output sound effect indicating that the commodity AA01 is put into the cart (sound effect output processing719).

The conversation proceeds to contents shown in aconversation display part721 inFIG. 30, and theuser10 issues an instruction to purchase the usual commodity which theuser10 often purchases (input722). In response to thisinput722, theinformation provision system1 references thepurchase history data272 and identifies the usual makeup remover (commodity AA02). At this time, theinformation provision system1 deems that theuser10 has attributes of preferring commodities which theuser10 gets used to using and thus, performs addition by a predetermined number to a score of the picky-about item “being picky about a product”. In addition, when due to this, the user type is changed, as described above, themember information data261 is updated.

In addition, theinformation provision system1 deems that with respect to theuser10, a preference degree related to said commodity is increased and performs addition to a preference degree score of said commodity.

Next, theinformation provision system1 confirms with theuser10 that the usual commodity is the commodity AA02 (response723).

Here, theuser10 confirms whether an economy size of the commodity AA02 is available (input724). Here, theinformation provision system1 references thecommodity data262 and checks whether or not the economy-size model of the commodity AA02 is available. In addition, theinformation provision system1 deems that theuser10 has attributes of being conscious about economical commodities and thus, performs addition by a predetermined number to a score of the picky-about item “being picky about a price”. In addition, when due to this, the user type is changed, as described above, themember information data261 is updated.

Since the economy-size model of the commodity AA02 is confirmed to be available, theinformation provision system1 generates and displays the fact and information pertinent to comparison with the commodity AA02 (response725). Here, when theuser10 makes a request as to a discount (input726), theinformation provision system1 deems that theuser10 has attributes of being conscious about economical commodities and thus, performs addition by a predetermined number to a score of the picky-about item “being picky about a price”. In addition, when due to this, the user type is changed, as described above, themember information data261 is updated.

The conversation proceeds to contents shown in aconversation display part731 inFIG. 31, and there, theinformation provision system1 conveys a message that the discount is impossible (response732). This information can be confirmed by referencing discount information pertinent to said commodity in thecommodity data262.

Next, theuser10 accepts the impossibility thereof and issues an instruction to purchase the economy-size model of the commodity AA02 (input733). In response to thisinput733, theinformation provision system1 makes aresponse734 saying “It is good to use plenty of it since a little burden is given to the skin, isn't it?” This is cited from, for example, recommendation information pertinent to said commodity AA02 (180 g) stored in thecommodity data262.

Theinformation provision system1 controls theuser terminal100 to output sound effect indicating that the commodity AA02 (180 g) is put into the cart (sound effect output processing735).

Thereafter, theuser10 makes aninput736 saying “Thanks you”. In response to this, theinformation provision system1 makes aresponse737 asking that there is other request. Here, while theuser10 is making aninput738 so as to recall other request (or after the elapse of a predetermined period of time since the response737), theinformation provision system1 determines that it is better to wait a response for a little while and controls a look of a character of the concierge to be changed so as to wait with the expression of expectation (look change processing739).

The conversation proceeds to contents shown in aconversation display part741 inFIG. 32, and theuser10 inputs a request related to a countermeasure against freckles (input742). Here, theinformation provision system1 deems that with respect to theuser10, a preference degree related to a freckle countermeasure commodity is increased and performs addition to a preference degree score of said commodity. The above-mentioned preference degree score can be recorded in thepreferred commodity data264.

Next, theinformation provision system1 makes a prompt to display a list of freckle-related commodities (response743). At this time, theinformation provision system1 references the purchase history of other user whose user type is the same as the user type of theuser10 and displays the list of freckle-related commodities based on predetermined criteria (for example, in the well-selling order) (display processing744). In addition, upon determining the recommended commodities as mentioned above, theinformation provision system1 references the purchase history and the commodity browse history of theuser10 and based on these, can determine the recommended commodities and the order of displaying the recommended commodities.

Here, further, theinformation provision system1 makes sure theuser10 has not purchased any freckle countermeasure commodity and makes an inquiry saying “Do you have any question?” (response745). In response thereto, theuser10 makes an input saying “I have no idea which one is good” (input746), and then, theinformation provision system1 proposes classifying the freckle-related commodities by purposes (response747). It is to be noted that in thecommodity data262, in addition to the commodity categories, by-purpose classifications are associated with the respective commodities.

Here, when theuser10 makes an input murmuring “I want to remove freckles” (input748), theinformation provision system1 determines that the input by theuser10 aims at “caring for freckles” and determines a commodity AA03 as a recommended commodity (response749). It is to be noted that here, theinformation provision system1 deems that with respect to theuser10, a preference degree related to caring for the freckles is increased and performs addition to a preference degree score of a preference target which is the caring for the freckles. The preference degree score as mentioned above can be recorded in thepreferred commodity data264.

The conversation proceeds to contents shown in aconversation display part751 inFIG. 33, and theuser10 makes a question related to the recommended commodity AA03, asking “Does this clear the freckles?” (input752). At this time, theinformation provision system1 deems that theuser10 has attributes of caring about effect and performs addition by a predetermined number to a score of a picky-about item “being picky about a product”. In addition, when due to this, the user type is changed, as described above, themember information data261 is updated.

Next, in response to the question of theuser10, theinformation provision system1 explains active ingredients and introduces comments of customers (response753). In addition, theinformation provision system1 controls the commodity display part corresponding to thecommodity display part126 to sequentially display the comments of customers in a pop-up manner together with this response753 (display processing754). It is to be noted that the explanation on the active ingredients and the comments of customers are extracted from detailed information of commodities and customer information stored in thecommodity data262 based on predetermined criteria.

Thereafter, when theuser10 makes aninput755 saying “It seems nice . . . ”, theinformation provision system1 determines based on thisinput755 that now, it is recommendation timing, extracts feeling in use from customer information of the commodity stored in thecommodity data262, allows theuser10 to feel a use image by displaying the extracted feeling in use, and provides a positive comment, which prompts theuser10 to purchase the commodity, for the user10 (response756).

Thereafter, when theuser10 makes aninput757 saying “But it's not now, I'll buy it on a payday!”, theinformation provision system1 determines that the recommendation timing is finished and causes aresponse758, which makes an offer of examining the commodity next time, to be displayed. In addition, here, theinformation provision system1 stores the commodity AA03 as a purchase examining commodity of theuser10 in thepreferred commodity data264.

The conversation proceeds to contents shown in aconversation display part761 inFIG. 34. Next, theuser10 asks a question as to a procedure upon moving home (input762). At this time, theinformation provision system1 collates contents of the question with theFAQ answer collection282 and causes a way of answering FAQ corresponding to the question (response763) to be displayed. Here, theinformation provision system1 prompts theuser10 to make notification by using an inquiry form or by a phone call and displays the inquiry form and a phone number in the commodity display part corresponding to the commodity display part126 (display processing764).

Then, theuser10 makes aninput765 saying “Well, I make a phone call.” and theinformation provision system1 makes aresponse766 saying “Yes, I am waiting for your call.”

Thereafter, when theuser10 conducts an address change procedure by the phone call (address change processing767) and makes aninput768 saying “The address change completed, thank you”, theinformation provision system1 determines that here, theuser10 intends to finish shopping and provides information as to free shipping in addition to a reply to the words of thanks from the user10 (response769).

The conversation proceeds to contents shown in aconversation display part771 inFIG. 35. Theuser10 asks a question as to a shipping cost (input772). Then, theinformation provision system1 collates contents of the question with theFAQ answer collection282 and causes a way of answering FAQ corresponding to the question (response773) to be displayed. Theinformation provision system1 calculates a shipping cost based on relationship between a total amount of prices of commodities in the cart and a shipping cost and notifies theuser10 of the calculated money amount of the shipping cost. In addition thereto, theinformation provision system1 notifies theuser10 of a money amount required for further shopping with which free shipping is applied and includes, in aresponse773, information that if the commodity AA03 which is the purchase examining commodity is purchased, the free shipping is applied.

Thereafter, when theuser10 makes aninput774 saying “Certainly, . . . what should I do”, theinformation provision system1 determines that now, it is recommendation timing and causes a seasonal recommendation sentence included in recommendation information related to the commodity AA03 and campaign information included in discount information stored in thecommodity data262 to be displayed as aresponse775.

Here, when theuser10 makes aninput776 saying “Hm, well, I'll buy it!”, theinformation provision system1, because theuser10 has accepted the recommended commodity, performs addition by a predetermined number to a score of the picky-about item “accepting recommendation”. In addition, when due to this, the user type is changed, as described above, themember information data261 is updated.

Base on the indication of the intention of the purchase by the user10 (input776), theinformation provision system1 causes words of thanks and a comment of recommendation from recommendation information related to the commodity AA03 (stored in the commodity data262) to be displayed (response777). Theinformation provision system1 shifts the conversation screen to a cart information confirmation screen for conducting a procedure of purchasing (display processing778). In addition, at this time, theinformation provision system1 stores the commodity AA03 in thepurchase history data272 so as to be associated with theuser10. In addition, success rates of recommended commodities may be separately registered.

Next, with reference toFIG. 36, a typical pattern of the above-described user serving processing in theinformation provision system1 according to the present embodiment will be described.FIG. 36 is a flowchart showing a typical procedure of the user serving processing.

First, at step S31, logging-in by theuser10 is accepted. A user ID and a password inputted from theuser terminal100 of the user10 (for example, from a top screen of Internet shopping) are authenticated based on themember information data261 in theinformation provision system1. When the authentication thereof is OK, the processing proceed to the next step, and when the authentication thereof is not OK, an error message is outputted onto theuser terminal100.

Next, at step S32, based on contents of utterance (input) by theuser10, extracted key words, and the like, theinformation provision system1 determines either to provide theuser10 with information pertinent to recommended commodities (that is, to make recommendation) or to wait for utterance by the user10 (that is, to make a hearing).

When it is determined that the hearing is made, the processing proceeds to step S33, theinformation provision system1 waits for the utterance by theuser10 there, and further, outputs a response so as to draw out utterance related to commodity preference of theuser10, thereby having a conversation with theuser10. Next, at step S34, theinformation provision system1 determines whether or not to continue the conversation with theuser10, and upon determining from contents of the utterance by theuser10 that theuser10 is not interested in a commodity or commodities or purchasing the commodity or the commodities (NO at step S34), the processing is finished.

Upon determining therefrom that theuser10 is interested in the commodity or commodities or purchasing the commodity or the commodities (YES at step S34), the processing returns to step S32 again, and the determination as to either the recommendation or the hearing is repeated.

At step S32, upon determining that the recommendation is made, the processing proceeds to step S35, and there, based on a user type of theuser10, contents of the utterance, a purchase history, and the like, a recommended commodity or commodities is or are determined. The determination of the recommended commodity or commodities is conducted, for example, in the procedure shown in the flowchart inFIG. 17. Next, at step S36, theinformation provision system1 controls the display of theuser terminal100 to display the determined recommended commodity (or a list of recommended commodities) (for example, causes the commodity display part corresponding to thecommodity display part126 to display the determined recommended commodity).

Next, at step S37, it is determined whether or not theuser10 is interested in the displayed recommended commodity or commodities, and when theuser10 is not interested in the displayed recommended commodity or commodities (NO at step S37), with the conversation being continues, the processing returns to step S32 again, and the determination as to either the recommendation or the hearing is repeated.

At step S37, when it is determined that theuser10 is interested in the displayed recommended commodity or commodities (YES at step S37), at step S38, further detailed information pertinent to the recommended commodity or commodities (for example, positive comments such as words of recommendation, campaign information, and comments of customers) is displayed, and theuser10 is prompted to click a commodity icon displayed in the commodity display part.

Next, at step S39, it is determined based on utterance and behavior of theuser10 whether or not theuser10 has intention of purchasing. At step S39, upon determining that theuser10 has no intention of purchasing (NO at step S39), with the conversation being continued, the processing returns to step S32 again, and the determination as to either the recommendation or the hearing is repeated.

Upon determining that theuser10 has the intention of purchasing (YES at step S39), at step S40, a flow of the procedure of purchasing is executed, and information related to a procedure required for purchasing the commodity is provided for theuser10.

Next, with reference toFIG. 37, in theinformation provision system1 according to the present embodiment, how to determine recommendation timing at which the recommended commodity is displayed and how to recommend a commodity will be described.FIG. 37 is a diagram showing several examples in which kinds of inputs byuser10, determination on the kinds of inputs, and final policies of recommendation based on the determination on the kinds of inputs are shown.

In a NO. 1 pattern inFIG. 37, when key words such as “So it is”, “It's true”, and “Certainly” are extracted from the input by theuser10, it is determined that theuser10 shows simple affirmation; and when key words such as “I'm not so sure” and “Not really” are extracted therefrom, it is determined that theuser10 shows simple negation.

In a NO. 2 pattern inFIG. 37, when key words such as “Aye”, “It is so”, “I know it”, and “it's nice” are extracted from the input by theuser10, it is determined that theuser10 shows interest; and when key words such as “Is there anything else?” are extracted therefrom, it is determined that theuser10 shows no interest.

In a NO. 3 pattern inFIG. 37, when an input by theuser10 is a comment “it is inorganic feeling, rather than cool one, isn't it”, such an input is determined to be a comment including evaluation items such as a design, a price, a function, usability, and reputation.

In a NO. 4 pattern inFIG. 37, when an input by theuser10 is a comment “Certainly, the design is good, but the price is . . . ”, it is determined that theuser10 shows composite negation. In addition, in a NO. 5 pattern, when an input by theuser10 is a comment “Is there any benefit for a member?”, the input is classified as other kind.

In the above-described NO. 1 to NO. 3 patterns, theinformation provision system1 controls the recommendation in accordance with a scenario so as to proceed to a flow in which that commodity or a commodity different from that commodity is recommended while probing whether theuser10 has purchasing willingness, as recommendation policy.

On the other hand, in the above-described NO. 4 pattern, when comments of affirmation and negation related to evaluation target items are inputted, theinformation provision system1 recommends a commodity by providing follow-up information pertinent to the targeted evaluation items and adding other evaluation items and controls the recommendation in accordance with a scenario so as to proceed to a flow in which that commodity or a commodity different from that commodity is recommended while probing whether theuser10 has purchasing willingness, as recommendation policy.

In addition, in the above-described NO. 5 pattern, on condition that there is no relation with the above-described patterns, when theuser10 makes an input or after that input, theinformation provision system1 makes a confirmation for that input and controls the recommendation in accordance with a scenario so as to introduce the next commodity or stand by as it is.

Next, with reference toFIG. 38, an example of a hardware configuration of a computer constituting theinformation provision server200 in theinformation provision system1 according to the embodiment of the present invention will be described. It is to be noted that the configuration of theinformation provision server200 shown inFIG. 38 merely exemplifies a representative configuration.

Theinformation provision server200 includes a CPU (Central Processing Unit)1001, a RAM (Random Access Memory)1002, a ROM (Read Only Memory)1003, anetwork interface1004, anaudio control part1005, amicrophone1006, aspeaker1007, adisplay controller1008, adisplay1009, aninput device interface1010, akeyboard1011, amouse1012, anexternal storage device1013, an externalrecording medium interface1014, and a bus1015 mutually connecting these constituent parts.

TheCPU1001 controls operation of the respective constituent parts in theinformation provision server200 and under the control of OS, controls execution of theinput analysis part202, theresponse control part203, and the like according to the present invention.

TheRAM1002 has temporarily stored therein programs for executing each processing executed by theCPU1001 and data used during executing each of these programs. TheROM1003 has stored therein programs and the like executed upon booting theinformation provision server200.

Thenetwork interface1004 is an interface for connecting to thenetwork1020. Thenetwork1020 is, for example, a network between theinformation provision server200 and theuser terminal100 and correspond to thenetwork300 shown inFIG. 1.

Theaudio control part1005 controls themicrophone1006 and thespeaker1007, controlling inputting and outputting of voice. Thedisplay controller1008 is a dedicated controller for actually processing drawing instructions issued by theCPU1001. Thedisplay1009 is a display device constituted of, for example, an LCD (Liquid Crystal Display) or a CRT (Cathode Ray Tube). Thedisplay1009 can also be constituted of a touch panel display which can be operated by touching.

Theinput device interface1010 receives signals inputted from thekeyboard1011 and themouse1012 and transmits, to theCPU1001, predetermined commands in accordance with signal patterns.

Theexternal storage device1013 is, for example, a storage device such as a hard disk and a semiconductor memory and has stored therein the above-mentioned programs and data, and upon executing the programs, as needed, the above-mentioned programs and data are loaded to theRAM1002. For example, the informationprovision management DB250 shown inFIG. 1 and the like can be stored in theexternal storage device1013.

The externalrecording medium interface1014 accesses theexternal recording medium1030 and reads data recorded therein. Theexternal recording medium1030 is, for example, a portable flash memory, a CD (Compact Disc), a DVD (Digital Versatile Disc), or the like. Programs, executed by theCPU1001, for realizing the respective functions of the present invention are provided via this externalrecording medium interface1014 from theexternal recording medium1030. In addition, as other distribution form in which the programs for realizing the respective functions of the present invention are provided, a route or the like in which the programs and the like are provided from a predetermined server on a network via thenetwork1020 and thenetwork interface1004 and stored in theexternal storage device1013 or theRAM1002 can also be considered.

The one example of the hardware configuration of theinformation provision server200 in theinformation provision system1 according to the one embodiment of the present invention is described. Basically, theuser terminal100 according to the present invention also is a computer having the same configuration as above. However, here, as for theinformation provision server200, theaudio control part1005, themicrophone1006, thespeaker1007, thedisplay controller1008, thedisplay1009, theinput device interface1010, thekeyboard1011, and themouse1012 are not indispensable constituent parts.

It is to be noted that although hereinabove, theinformation provision system1 according to the one embodiment of the present invention is described by illustrating the specific examples which allow the present invention to be implemented, each of these specific examples is merely one example for describing the present invention, and the scope of the right of the present invention is not limited to these specific examples. By employing various other methods and configurations, the technical idea of the present invention can be realized.

REFERENCE SIGNS LIST

- 1 information provision system
- 10 user
- 100 user terminal
- 200 information provision server
- 250 information provision management DB
- 300 network
- 1001 CPU
- 1002 RAM
- 1003 ROM
- 1004 network interface
- 1005 audio control part
- 1006 microphone
- 1007 speaker
- 1008 display controller
- 1009 display
- 1010 input device interface
- 1011 keyboard
- 1012 mouse
- 1013 external storage device
- 1014 external recording medium interface
- 1015 bus

Claims

What is claimed is:

1. An information provision system comprising:

the response determination device controlling information pertinent to the determined commodity or service so as to be included in the contents of the response.

2. The information provision system according toclaim 1, wherein based on the determined user type of the user and a category selected by the user on the user terminal, the response determination device determines the commodity or the service to be presented to the user.

3. The information provision system according toclaim 1, wherein based on purchase information related to a user whose user type is the same as or similar to the determined user type of the user, the response determination device determines the commodity or the service to be presented to the user.

4. The information provision system according toclaim 1, wherein based on the words obtained based on the input and attributes of the user, the user type determination device determines the user type of the user.

5. The information provision system according toclaim 1, wherein in accordance with the words received over time, the user type determination device updates the user type of the user.

6. An information provision method implemented on an information provision server in an information provision system, the method comprising:

a user type determination step of determining a user type of the user by using words obtained based on the input, wherein

at the response determination step, based on the determined user type of the user, the commodity or the service to be presented to the user is determined, and information pertinent to the determined commodity or service is controlled so as to be included in the contents of the response.

7. A non-transitory computer-readable storage medium having stored therein a program for causing a computer to function as: