CN111914077A - Customized speech recommendation method, device, computer equipment and storage medium - Google Patents
Customized speech recommendation method, device, computer equipment and storage mediumDownload PDFInfo
- Publication number
- CN111914077A CN111914077ACN202010784769.2ACN202010784769ACN111914077ACN 111914077 ACN111914077 ACN 111914077ACN 202010784769 ACN202010784769 ACN 202010784769ACN 111914077 ACN111914077 ACN 111914077A
- Authority
- CN
- China
- Prior art keywords
- data
- model
- recommendation
- training
- customized
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/335—Filtering based on additional data, e.g. user or group profiles
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0281—Customer communication at a business location, e.g. providing product or service information, consulting
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Artificial Intelligence (AREA)
- Strategic Management (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Human Computer Interaction (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Description
Technical Field
The invention relates to the technical field of intelligent customer service, in particular to a customized speech recommendation method, a customized speech recommendation device, computer equipment and a storage medium.
Background
The insurance customer service is one of the services closest to the customers in the insurance industry, not only relates to the propaganda of insurance knowledge, the expansion of insurance business, the consultation before and after sale and the like, but also directly drives the business volume and the achievement of an insurance company, so that the excellent customer service can continuously attract more customers and improve the insurance service experience of the customers. With the development of artificial intelligence technology and natural language technology, more and more companies adopt robots to provide customer service question-answering services.
In the prior art, the robot customer service adopts the traditional intelligent outbound technology, although the voice information of the client can be obtained in real time and the intention of the client can be understood, the personal characteristics, interests and hobbies of the client, the demand degree of marketing products and the like can not be known generally, so that the reply of the robot customer service is uniform regardless of the identity information and the personal hobbies of the client, and corresponding marketing strategies can not be formulated for different clients.
Disclosure of Invention
The embodiment of the disclosure provides a customized tactical recommendation method, a customized tactical recommendation device, computer equipment and a storage medium. The following presents a simplified summary in order to provide a basic understanding of some aspects of the disclosed embodiments. This summary is not an extensive overview and is intended to neither identify key/critical elements nor delineate the scope of such embodiments. Its sole purpose is to present some concepts in a simplified form as a prelude to the more detailed description that is presented later.
In a first aspect, an embodiment of the present disclosure provides a customized speech recommendation method, including:
acquiring first data for representing user characteristics and second data of a user question;
respectively inputting the first data and the second data into a preset recommendation model to obtain product data matched with user preference;
and inputting the first data, the second data and the product data into a preset phonetics model to generate a customized phonetics text.
In one embodiment, before the first data and the second data are respectively input into the preset recommendation model, the method further includes:
constructing a first training sample set according to the first data and the second data;
and training a recommendation model based on the first training sample set and the deep FM recommendation algorithm.
In one embodiment, training the recommendation model based on the first set of training samples and the deep fm recommendation algorithm includes:
and training the recommendation model by adopting a multi-task collaborative training mode.
In one embodiment, before inputting the first data, the second data and the product data into the preset conversational model, the method further comprises:
constructing a second training sample set according to the first data, the second data and the product data;
training the conversational model based on the second set of training samples and the Transformer model.
In one embodiment, further comprising: and training the recommendation model and the dialect model in a joint training mode.
In one embodiment, training the recommendation model and the conversational model by means of joint training comprises:
inputting the deep representation of the data of the recommendation model hidden layer into a speaking model;
the embedding vector of the dialogistic model is input into the recommendation model.
In one embodiment, after generating the customized phonemic text, further comprising:
customized verbal text is entered into the intelligent outbound system.
In a second aspect, an embodiment of the present disclosure provides a customized tactical recommendation apparatus, including:
the system comprises an acquisition module, a processing module and a processing module, wherein the acquisition module is used for acquiring first data for representing user characteristics and second data of a user question;
the recommendation module is used for respectively inputting the first data and the second data into a preset recommendation model to obtain product data matched with the user preference degree;
and the language-operation module is used for inputting the first data, the second data and the product data into a preset language-operation model to generate a customized language-operation text.
In a third aspect, the disclosed embodiments provide a computer device, including a memory and a processor, where the memory stores computer readable instructions, and the computer readable instructions, when executed by the processor, cause the processor to perform the steps of the customized talk recommendation method provided in the above embodiments.
In a fourth aspect, embodiments of the present disclosure provide a storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the customized dialog recommendation method provided by the above-described embodiments.
The technical scheme provided by the embodiment of the disclosure can have the following beneficial effects:
according to the customized language recommendation method provided by the embodiment of the disclosure, the first data representing the characteristics of the user and the second data representing the problems of the user can be combined to generate the product data matched with the preference degree of the user, the product matched with the interest of the user can be recommended for different users, and the customized language can be automatically generated.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and together with the description, serve to explain the principles of the invention.
FIG. 1 is a diagram of an implementation environment for a method of customizing a conversational recommendation, according to an example embodiment;
FIG. 2 is a diagram illustrating an internal structure of a computer device in accordance with one illustrative embodiment;
FIG. 3 is a flow diagram illustrating a method of customizing a conversational recommendation, according to an example embodiment;
FIG. 4 is a flow diagram illustrating a method of customizing a conversational recommendation, according to an example embodiment;
FIG. 5 is a diagram illustrating a method of customizing a conversational recommendation, according to an example embodiment;
FIG. 6 is a block diagram illustrating a customized dialog recommendation device, according to an exemplary embodiment;
FIG. 7 is a block diagram illustrating a customized tactical recommendation device, according to an exemplary embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
It will be understood that, as used herein, the terms "first," "second," and the like may be used herein to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish one element from another. For example, a first field and algorithm determination module may be referred to as a second field and algorithm determination module, and similarly, a second field and algorithm determination module may be referred to as a first field and algorithm determination module, without departing from the scope of the present application.
Fig. 1 is a diagram illustrating an implementation environment of a customized talk recommendation method according to an exemplary embodiment, as shown in fig. 1, in which aserver 110 and aterminal 120 are included.
Theserver 110 is a customized speech recommendation device, such as a computer device used by a technician, and theserver 110 has a customized speech recommendation tool installed thereon. Theterminal 120 is installed with an application that needs to perform customized speech recommendation, when a customized speech needs to be recommended, a technician may send a request for recommending the customized speech at thecomputer device 110, where the request carries a request identifier, and thecomputer device 110 receives the request to obtain data information representing user characteristics stored in thecomputer device 110. And then generating a customized phonetics text by using the preset recommendation model and the preset phonetics model, and transmitting the customized phonetics text to the intelligent calling system on theterminal 120 to communicate with the user.
It should be noted that theterminal 120 and thecomputer device 110 may be, but are not limited to, a smart phone, a tablet computer, a notebook computer, a desktop computer, and the like. Thecomputer device 110 and theterminal 120 may be connected through bluetooth, USB (Universal Serial Bus), or other communication connection methods, which is not limited herein.
FIG. 2 is a diagram illustrating an internal structure of a computer device according to an exemplary embodiment. As shown in fig. 2, the computer device includes a processor, a non-volatile storage medium, a memory, and a network interface connected through a system bus. The non-volatile storage medium of the computer device stores an operating system, a database and computer readable instructions, the database can store control information sequences, and the computer readable instructions when executed by the processor can enable the processor to realize a customized tactical recommendation method. The processor of the computer device is used for providing calculation and control capability and supporting the operation of the whole computer device. The memory of the computer device may have stored therein computer readable instructions that, when executed by the processor, may cause the processor to perform a method of customizing a speech recommendation. The network interface of the computer device is used for connecting and communicating with the terminal. Those skilled in the art will appreciate that the architecture shown in fig. 2 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
The customized literature recommendation method provided by the embodiments of the present application will be described in detail below with reference to fig. 3-5. The method may be implemented in dependence on a computer program, operable on a data transmission device based on the von neumann architecture. The computer program may be integrated into the application or may run as a separate tool-like application.
Referring to fig. 3, a schematic flow chart of a customized speech recommendation method is provided in an embodiment of the present application, and as shown in fig. 3, the method in the embodiment of the present application may include the following steps:
s301, first data used for representing user characteristics and second data of user problems are obtained.
With the rapid development of computer technology, the marketing mode of the traditional insurance industry is gradually overturned, and in a typical telemarketing scene, the personnel in the seat usually cannot know the personal characteristics, the interests and hobbies of the client, the demand degree of the product to be marketed and the like. The blindly mass marketing is like a large sea fishing needle, and the advantages of the operation and the resources cannot be fully utilized.
Through the years of telemarketing and the accumulation of data information of online and offline sales, the insurance group can collect a large amount of evaluation information and feedback information of customers on marketing products. Meanwhile, the self information of the client can be stored in the system along with the signing of the insurance contract. By utilizing the data, independent marketing methods can be made for different customers, the customers are automatically marketed through the intelligent outbound system, and the privacy of the data can be guaranteed because the dialing of real seats is not involved.
Specifically, first data characterizing the user characteristics and second data of the user question stored in the system may be obtained, and in some possible implementations, the first data includes one or more of name information, age information, gender information, family member information, interested product information, and product demand level information of the user. The second data includes user feedback information on the purchased product and information on the product to be marketed.
By combining the information of the user and the information of the product to be marketed, the interested marketing product is pushed for the client, and the advantages of information resources are fully utilized.
S302, the first data and the second data are respectively input into a preset recommendation model, and product data matched with the user preference degree are obtained.
Further, before the first data and the second data are input into the preset recommendation model, training the recommendation model is further included.
Specifically, data screening is performed on the acquired first data and second data, data which is purchased by a product and has forward evaluation is recorded as an effective sample, and preprocessing is performed on the effective sample obtained after screening, including deleting or filling up abnormal data and missing data, so as to obtain corrected data.
And constructing a first training sample set according to the screened and preprocessed first data and second data, and training a recommendation model based on the constructed first training sample set and a recommendation algorithm. The common recommendation algorithms comprise an LR + GBDT recommendation algorithm, GBDT can be used for learning high-order nonlinear feature combination, LR is used for learning ID features, and the fusion has good effect, and a DeepFM recommendation algorithm, which uses a dense vector fusion factorization machine and a neural network to extract first-order features, uses the neural network to extract high-order features, and has good technical effect.
In some possible implementation manners, a recommendation model is trained based on a first training sample set and a deep fm recommendation algorithm, the recommendation model provided in the embodiment of the disclosure accesses deep characterization of data obtained by a hidden layer of the recommendation model into a dialectic model, and a multi-task collaborative training manner is adopted for training. The first task is to sequence the recommended tasks, and the loss function adopted is as follows:
wherein,
the deep characterization of the data obtained for the hidden layer,
a deep characterization of the data obtained for the positive sample,
and f can be used as a calculation process of a deep FM algorithm for the deep characterization of the data obtained by the negative sample.
The second task is the return transmission of the phonetics generation task of the phonetics model, and the adopted loss function is as follows:
wherein, yi,kRepresenting each word in the real recommended dialect, pi,kRepresenting the probabilities predicted by all words in the lexicon, the objective of the loss function is to make the resulting dialect as close as possible to the real one.
According to the method, the trained recommendation model is obtained, and the main body of the recommendation model is a recommendation model based on the content. And when performing conversational recommendation, inputting the first data and the second data corresponding to the user to be marketed into a recommendation model, and obtaining product data matched with the user preference according to the personal information of the user.
According to the steps, products matched with the interests of the users can be recommended for different users, and therefore the problem that different products cannot be marketed according to different users in the prior art is solved.
S303, inputting the first data, the second data and the product data into a preset phonetics model to generate a customized phonetics text.
Further, before inputting the first data, the second data and the product data into the dialogistic model, training the dialogistic model is further included.
Specifically, a second training sample set is constructed from the first data, the second data and the product data, and the conversational model is trained based on the second training sample set and the conversational algorithm. Commonly used dialoging algorithms include RNN algorithms and LSTM algorithms, both of which use text sequences as training samples to learn word-to-word associations, thereby yielding the ability to generate dialogies.
In some possible implementation manners, the jargon model can be trained through a transformer model, the transformer model breaks through the limitation that the RNN model cannot calculate in parallel, the operation number required for calculating the association between two positions does not increase along with the distance, and more importantly, the transformer model applies an attention mechanism to better focus on important contents in the text.
The conversational model is trained by a task-independent generation method, and the loss function of the conversational model is the same as the loss function in step S302, and is:
wherein, yi,kRepresenting each word in the real recommended dialect, pi,kRepresenting the probabilities predicted by all words in the lexicon, the objective of the loss function is to make the resulting dialect as close as possible to the real one.
In some possible implementations, the conversational model and the recommendation model generate the customized conversational language by a joint training and joint prediction mode. Specifically, in the training process of the recommendation model, an embedded vector in the dialect model is introduced as an additional feature, and in the training process of the dialect model, a vector of a high-dimensional hidden layer of the recommendation model is introduced as a representation of a client and a product to be recommended, so that the generated dialect can be adjusted in a self-adaptive manner and better accords with a real dialogue scene.
And according to the trained speech model, a customized speech text can be obtained, and the generated customized speech can improve the interest of the user and has stronger attraction to the client.
Further, after generating the customized phonemic text, the customized phonemic text may be entered into the intelligent calling system.
The intelligent outbound system is a dialogue system for automatically recognizing and generating voice, and in some possible implementation modes, the intelligent outbound system is an intelligent telemarketing robot which generates outbound opening white according to received customized dialogue texts, and compared with the previous uniform opening white, the intelligent outbound opening white is more attractive to customers, and meanwhile, different products are recommended for the customers under different conditions to achieve the customized effect.
In some exemplary scenes, the intelligent telemarketing robot knows that the name of the user to be marketed is somebody in forest, sex male, interest and hobby are driving, and the demand degree on the car insurance is high according to the obtained customized dialect. Therefore, the calling-out field of the robot can be 'good for Mr. Lin, i is the safe of safe insurance, and the vehicle insurance of our company is recently acted, so that people can know the position conveniently'.
In some exemplary scenarios, the intelligent telemarketing robot knows that the name of the user to be marketed is somebody plum and sex male, the interest is financing, and the demand degree for financing is high according to the obtained customized terminology. Therefore, the out-calling field of the robot can be 'good for Mr. Li, i is safe for safety, and the financial affairs of our company are recently done to facilitate understanding'.
According to the steps, the intelligent outbound system can communicate with the user according to the received customized language text, provides interesting products for the user, realizes accurate marketing, greatly improves the purchase rate of the products, and can arouse the interest of the user through the customized language.
Optionally, the customized dialect recommendation method provided by the embodiment of the present disclosure may be applied to other industries such as an insurance industry, a house sales industry, a course sales industry, and the like, and the embodiment of the present disclosure is not particularly limited.
To facilitate understanding of the customized dialogies recommendation provided by the embodiments of the present application, reference is made to fig. 4. As shown in fig. 4, a customized dialog recommendation method includes:
s401, first data used for representing user characteristics and second data of the user problems are obtained.
S402, constructing a first training sample set according to the first data and the second data.
S403, training the recommendation model based on the first training sample set and the DeepFM recommendation algorithm.
S404, inputting the first data and the second data into a preset recommendation model respectively to obtain product data matched with user preference.
S405, constructing a second training sample set according to the first data, the second data and the product data.
S406, training the speaking model based on the second training sample set and the Transformer model.
S407, inputting the first data, the second data and the product data into a preset phonetics model to generate a customized phonetics text.
To facilitate understanding of the customized literature recommendation provided by embodiments of the present application, reference is made to fig. 5. As shown in fig. 5, a customized conversational recommendation method, which uses a hierarchical modeling method, includes:
the data layer stores a large amount of customer information including name information, age information, gender information, family address information, family member information, interesting product information and the like of customers, stores the information of purchased products of the customers, and also stores the information of products to be marketed, and the evaluation feedback information of the products. And screening and preprocessing the data in the data layer to obtain first data and second data representing the user characteristics. The first data and the second data are input into the model layer.
And the model layer comprises a recommendation model and a conversational model, and the recommendation model and the conversational model are trained according to the received first data and second data, and the conversational model and the recommendation model adopt a joint training and joint prediction mode to generate a customized conversational operation. Specifically, in the training process of the recommendation model, an embedded vector in the dialect model is introduced as an extra feature, and in the training process of the dialect model, a vector of a high-dimensional hidden layer of the recommendation model is introduced as the representation of a client and a product to be recommended, so that the generated dialect can be adaptively adjusted. And inputting the obtained dialect text into the application layer.
And the application layer inputs the received dialect text into the intelligent outbound system to generate a dialogue, communicates with the user and recommends an interested product for the user.
The customized dialect recommendation method provided by the embodiment of the disclosure can recommend products matched with the interests of users for different users, and automatically generate customized dialects, and the generated customized dialects can improve the interests of the users and have stronger attractiveness to the customers, so that accurate marketing is realized, and the purchase rate of the products is improved.
The following are embodiments of the apparatus of the present invention that may be used to perform embodiments of the method of the present invention. For details which are not disclosed in the embodiments of the apparatus of the present invention, reference is made to the embodiments of the method of the present invention.
Referring to fig. 6, a schematic structural diagram of a customized tactical recommendation apparatus according to an exemplary embodiment of the present invention is shown. As shown in fig. 6, the customized tactical recommendation apparatus may be integrated in thecomputer device 110, and specifically may include an obtainingmodule 601, a recommendingmodule 602, and atactical module 603.
An obtainingmodule 601, configured to obtain first data used for characterizing a user characteristic and second data of a user question;
therecommendation module 602 is configured to input the first data and the second data into a preset recommendation model respectively to obtain product data matched with the user preference;
thephonetics module 603 is configured to input the first data, the second data, and the product data into a preset phonetics model, and generate a customized phonetics text.
In one embodiment, as shown in fig. 7, the customized tactical recommendation apparatus may further include afirst training module 604 for constructing a first training sample set from the first data and the second data for training the recommendation model based on the first training sample set and the deep fm recommendation algorithm.
In one embodiment, thefirst training module 604 is further configured to train the recommendation model in a multi-task collaborative training manner.
In one embodiment, the customized dialog recommendation device may further include asecond training module 605 for constructing a second training sample set from the first data, the second data, and the product data for training the dialog model based on the second training sample set and the Transformer model.
In one embodiment, thefirst training module 604 and thesecond training module 605 are also used to train the recommendation model and the conversational model by way of joint training.
In one embodiment, thefirst training module 604 and thesecond training module 605 are specifically configured to input deep representations of data of the recommendation model hidden layers into the conversational model; the embedding vector of the dialogistic model is input into the recommendation model.
In one embodiment, the customized phonetics recommendation device may also include an intelligentoutbound module 606 for entering customized phonetics text into the intelligent outbound system.
Based on the customized language recommendation device provided by the embodiment of the disclosure, products matched with the interests of users can be recommended for different users, and customized languages are automatically generated.
It should be noted that, when the customized speech recommendation apparatus provided in the foregoing embodiment executes the customized speech recommendation method, only the division of the functional modules is illustrated, and in practical applications, the function distribution may be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules, so as to complete all or part of the functions described above. In addition, the customized literature recommendation device provided in the above embodiments and the customized literature recommendation method embodiment belong to the same concept, and the implementation process is detailed in the method embodiment, which is not described herein again.
In one embodiment, a computer device is proposed, the computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the following steps when executing the computer program: acquiring first data for representing user characteristics and second data of a user question; respectively inputting the first data and the second data into a preset recommendation model to obtain product data matched with user preference; and inputting the first data, the second data and the product data into a preset phonetics model to generate a customized phonetics text.
In one embodiment, the processor, when executing the computer readable instructions, further performs the steps of: before the first data and the second data are respectively input into the preset recommendation model, the method further comprises the following steps: constructing a first training sample set according to the first data and the second data; and training a recommendation model based on the first training sample set and the deep FM recommendation algorithm.
In one embodiment, the processor, when executing the computer readable instructions, further performs the steps of: and training the recommendation model by adopting a multi-task collaborative training mode.
In one embodiment, the processor, when executing the computer readable instructions, further performs the steps of: before the first data, the second data and the product data are all input into the preset conversational model, the method further comprises the following steps: constructing a second training sample set according to the first data, the second data and the product data; training the conversational model based on the second set of training samples and the Transformer model.
In one embodiment, the processor, when executing the computer readable instructions, further performs the steps of: and training the recommendation model and the dialect model in a joint training mode.
In one embodiment, when training the recommendation model and the conversational model by joint training, the processor when executing the computer readable instructions further performs the steps of: inputting the deep representation of the data of the recommendation model hidden layer into a speaking model; the embedding vector of the dialogistic model is input into the recommendation model.
In one embodiment, the processor, when executing the computer readable instructions, further performs the steps of: customized verbal text is entered into the intelligent outbound system.
In one embodiment, a storage medium is provided that stores computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps of: acquiring first data for representing user characteristics and second data of a user question; respectively inputting the first data and the second data into a preset recommendation model to obtain product data matched with user preference; and inputting the first data, the second data and the product data into a preset phonetics model to generate a customized phonetics text.
In one embodiment, the processor, when executing the computer readable instructions, further performs the steps of: before the first data and the second data are respectively input into the preset recommendation model, the method further comprises the following steps: constructing a first training sample set according to the first data and the second data; and training a recommendation model based on the first training sample set and the deep FM recommendation algorithm.
In one embodiment, the processor, when executing the computer readable instructions, further performs the steps of: and training the recommendation model by adopting a multi-task collaborative training mode.
In one embodiment, the processor, when executing the computer readable instructions, further performs the steps of: before the first data, the second data and the product data are all input into the preset conversational model, the method further comprises the following steps: constructing a second training sample set according to the first data, the second data and the product data; training the conversational model based on the second set of training samples and the Transformer model.
In one embodiment, the processor, when executing the computer readable instructions, further performs the steps of: and training the recommendation model and the dialect model in a joint training mode.
In one embodiment, when training the recommendation model and the conversational model by joint training, the processor when executing the computer readable instructions further performs the steps of: inputting the deep representation of the data of the recommendation model hidden layer into a speaking model; the embedding vector of the dialogistic model is input into the recommendation model.
In one embodiment, the processor, when executing the computer readable instructions, further performs the steps of: customized verbal text is entered into the intelligent outbound system.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a computer-readable storage medium, and can include the processes of the embodiments of the methods described above when the computer program is executed. The storage medium may be a non-volatile storage medium such as a magnetic disk, an optical disk, a Read-Only Memory (ROM), or a Random Access Memory (RAM).
The technical features of the embodiments described above may be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the embodiments described above are not described, but should be considered as being within the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (10)
1. A method for customized conversational recommendation, comprising:
acquiring first data for characterizing user features and second data of the user questions;
inputting the first data and the second data into a preset recommendation model respectively to obtain product data matched with user preference;
and inputting the first data, the second data and the product data into a preset phonetics model to generate a customized phonetics text.
2. The method of claim 1, wherein before inputting the first data and the second data into the preset recommendation model, respectively, further comprising:
constructing a first training sample set according to the first data and the second data;
training the recommendation model based on the first training sample set and a deep FM recommendation algorithm.
3. The method of claim 2, wherein training the recommendation model based on the first set of training samples and a deep fm recommendation algorithm comprises:
and training the recommendation model in a multi-task collaborative training mode.
4. The method of claim 1, wherein prior to inputting the first data, the second data, and the product data into a preset conversational model, further comprising:
constructing a second training sample set according to the first data, the second data and the product data;
training the conversational model based on the second set of training samples and the Transformer model.
5. The method according to any one of claims 2-4, further comprising:
and training the recommendation model and the dialect model in a joint training mode.
6. The method of claim 5, wherein the training the recommendation model and the conversational model by way of joint training comprises:
inputting the data deep representation of the recommendation model hidden layer into the phonetics model;
and inputting the embedding vector of the dialogical model into the recommended model.
7. The method of claim 1, wherein after generating the customized verbal text, further comprising:
inputting the customized dialect text into an intelligent outbound system.
8. A customized dialog recommendation device, comprising:
the acquisition module is used for acquiring first data for representing user characteristics and second data of the user problems;
the recommendation module is used for respectively inputting the first data and the second data into a preset recommendation model to obtain product data matched with user preference;
and the phonetics module is used for inputting the first data, the second data and the product data into a preset phonetics model to generate a customized phonetics text.
9. A computer device comprising a memory and a processor, the memory having stored therein computer readable instructions which, when executed by the processor, cause the processor to perform the steps of the method of customizing a speech recommendation according to any one of claims 1 to 7.
10. A storage medium having stored thereon computer-readable instructions which, when executed by one or more processors, cause the one or more processors to perform the steps of the method of customizing a word-of-speech recommendations according to any one of claims 1 to 7.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010784769.2ACN111914077A (en) | 2020-08-06 | 2020-08-06 | Customized speech recommendation method, device, computer equipment and storage medium |
| PCT/CN2020/119370WO2021139281A1 (en) | 2020-08-06 | 2020-09-30 | Customized speech skill recommendation method and apparatus, computer device, and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010784769.2ACN111914077A (en) | 2020-08-06 | 2020-08-06 | Customized speech recommendation method, device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111914077Atrue CN111914077A (en) | 2020-11-10 |
Family
ID=73286529
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010784769.2APendingCN111914077A (en) | 2020-08-06 | 2020-08-06 | Customized speech recommendation method, device, computer equipment and storage medium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN111914077A (en) |
| WO (1) | WO2021139281A1 (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112651569A (en)* | 2020-12-31 | 2021-04-13 | 中国平安财产保险股份有限公司 | Product recommendation method, device and equipment based on state nodes and storage medium |
| CN113111254A (en)* | 2021-03-08 | 2021-07-13 | 支付宝(杭州)信息技术有限公司 | Training method, fitting method and device of recommendation model and electronic equipment |
| CN113806517A (en)* | 2021-09-22 | 2021-12-17 | 未鲲(上海)科技服务有限公司 | An outbound call method, device, equipment and medium based on machine learning model |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109146610A (en)* | 2018-07-16 | 2019-01-04 | 众安在线财产保险股份有限公司 | It is a kind of intelligently to insure recommended method, device and intelligence insurance robot device |
| CN109522483A (en)* | 2018-11-14 | 2019-03-26 | 北京百度网讯科技有限公司 | Method and apparatus for pushed information |
| CN110032630A (en)* | 2019-03-12 | 2019-07-19 | 阿里巴巴集团控股有限公司 | Talk about art recommendation apparatus, method and model training equipment |
| CN110175012A (en)* | 2019-04-17 | 2019-08-27 | 百度在线网络技术（北京）有限公司 | Technical ability recommended method, device, equipment and computer readable storage medium |
| CN110209774A (en)* | 2018-02-11 | 2019-09-06 | 北京三星通信技术研究有限公司 | Handle the method, apparatus and terminal device of session information |
| CN110263244A (en)* | 2019-02-14 | 2019-09-20 | 腾讯科技（深圳）有限公司 | Content recommendation method, device, storage medium and computer equipment |
| CN110738545A (en)* | 2019-08-30 | 2020-01-31 | 深圳壹账通智能科技有限公司 | Product recommendation method and device based on user intention identification, computer equipment and storage medium |
| CN110929162A (en)* | 2019-12-04 | 2020-03-27 | 腾讯科技（深圳）有限公司 | Recommendation method and device based on interest points, computer equipment and storage medium |
| CN111079015A (en)* | 2019-12-17 | 2020-04-28 | 腾讯科技（深圳）有限公司 | Recommendation method and device, computer equipment and storage medium |
| CN111144937A (en)* | 2019-12-20 | 2020-05-12 | 北京达佳互联信息技术有限公司 | Advertisement material determination method, device, equipment and storage medium |
| CN111259132A (en)* | 2020-01-16 | 2020-06-09 | 中国平安财产保险股份有限公司 | Method, apparatus, computer equipment and storage medium for speech recommendation |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10565655B2 (en)* | 2017-05-16 | 2020-02-18 | Capital One Services, Llc | Systems and methods for providing customized financial advice |
| CN110459210A (en)* | 2019-07-30 | 2019-11-15 | 平安科技（深圳）有限公司 | Answering method, device, equipment and storage medium based on speech analysis |
| CN110555100A (en)* | 2019-09-06 | 2019-12-10 | 北京讯鸟软件有限公司 | Multi-product demand matching method and system based on graph generation free dialogue |
- 2020
- 2020-08-06CNCN202010784769.2Apatent/CN111914077A/enactivePending
- 2020-09-30WOPCT/CN2020/119370patent/WO2021139281A1/ennot_activeCeased
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110209774A (en)* | 2018-02-11 | 2019-09-06 | 北京三星通信技术研究有限公司 | Handle the method, apparatus and terminal device of session information |
| CN109146610A (en)* | 2018-07-16 | 2019-01-04 | 众安在线财产保险股份有限公司 | It is a kind of intelligently to insure recommended method, device and intelligence insurance robot device |
| CN109522483A (en)* | 2018-11-14 | 2019-03-26 | 北京百度网讯科技有限公司 | Method and apparatus for pushed information |
| CN110263244A (en)* | 2019-02-14 | 2019-09-20 | 腾讯科技（深圳）有限公司 | Content recommendation method, device, storage medium and computer equipment |
| CN110032630A (en)* | 2019-03-12 | 2019-07-19 | 阿里巴巴集团控股有限公司 | Talk about art recommendation apparatus, method and model training equipment |
| CN110175012A (en)* | 2019-04-17 | 2019-08-27 | 百度在线网络技术（北京）有限公司 | Technical ability recommended method, device, equipment and computer readable storage medium |
| CN110738545A (en)* | 2019-08-30 | 2020-01-31 | 深圳壹账通智能科技有限公司 | Product recommendation method and device based on user intention identification, computer equipment and storage medium |
| CN110929162A (en)* | 2019-12-04 | 2020-03-27 | 腾讯科技（深圳）有限公司 | Recommendation method and device based on interest points, computer equipment and storage medium |
| CN111079015A (en)* | 2019-12-17 | 2020-04-28 | 腾讯科技（深圳）有限公司 | Recommendation method and device, computer equipment and storage medium |
| CN111144937A (en)* | 2019-12-20 | 2020-05-12 | 北京达佳互联信息技术有限公司 | Advertisement material determination method, device, equipment and storage medium |
| CN111259132A (en)* | 2020-01-16 | 2020-06-09 | 中国平安财产保险股份有限公司 | Method, apparatus, computer equipment and storage medium for speech recommendation |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112651569A (en)* | 2020-12-31 | 2021-04-13 | 中国平安财产保险股份有限公司 | Product recommendation method, device and equipment based on state nodes and storage medium |
| CN113111254A (en)* | 2021-03-08 | 2021-07-13 | 支付宝(杭州)信息技术有限公司 | Training method, fitting method and device of recommendation model and electronic equipment |
| CN113111254B (en)* | 2021-03-08 | 2023-04-07 | 支付宝(杭州)信息技术有限公司 | Training method, fitting method and device of recommendation model and electronic equipment |
| CN113806517A (en)* | 2021-09-22 | 2021-12-17 | 未鲲(上海)科技服务有限公司 | An outbound call method, device, equipment and medium based on machine learning model |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2021139281A1 (en) | 2021-07-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107846350B (en) | Method, computer readable medium and system for context-aware network chat | |
| US11625425B2 (en) | Dialogue management system with hierarchical classification and progression | |
| CN110070391B (en) | Data processing method and device, computer readable medium and electronic equipment | |
| US9177318B2 (en) | Method and apparatus for customizing conversation agents based on user characteristics using a relevance score for automatic statements, and a response prediction function | |
| US9875494B2 (en) | Using intents to analyze and personalize a user's dialog experience with a virtual personal assistant | |
| CN110738545A (en) | Product recommendation method and device based on user intention identification, computer equipment and storage medium | |
| US20200051451A1 (en) | Short answer grade prediction | |
| CN111914077A (en) | Customized speech recommendation method, device, computer equipment and storage medium | |
| CN113112282A (en) | Method, device, equipment and medium for processing consult problem based on client portrait | |
| US20150142706A1 (en) | Systems and Methods for Creating and Implementing an Artificially Intelligent Agent or System | |
| CN115062627B (en) | Method and apparatus for an artificial intelligence based computer aided persuasion system | |
| CN118193754A (en) | Task execution method, device, equipment, medium and program product | |
| Uchida et al. | Female-type android’s drive to quickly understand a user’s concept of preferences stimulates dialogue satisfaction: Dialogue strategies for modeling user’s concept of preferences | |
| US20240330654A1 (en) | Generative Artificial Intelligence as a Personal Task Generator to Complete Objectives | |
| Aattouri et al. | Modeling of an artificial intelligence based enterprise callbot with natural language processing and machine learning algorithms | |
| CN111046151A (en) | Message processing method and device | |
| CN114296547B (en) | Method, device and storage medium for initiating active dialogue | |
| CN114239604A (en) | Online consultation processing method, device and computer equipment | |
| CN119127648A (en) | Intelligent agent evaluation method, device, electronic device and storage medium | |
| US20240338710A1 (en) | Real-time assistance for a customer at a point of decision through hardware and software smart indicators deterministically generated through artificial intelligence | |
| CN111651582B (en) | Method and system for simulating user speaking | |
| CN116628179B (en) | User operation data visualization and man-machine interaction recommendation method | |
| CN116956856A (en) | Data processing method and device, storage medium and electronic equipment | |
| JP2021026686A (en) | Character display device, character display method, and program | |
| CN112115717A (en) | Data processing method, device and equipment and readable storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | Application publication date:20201110 | |
| RJ01 | Rejection of invention patent application after publication |