CN112562658A - Groove filling method and device - Google Patents

Abstract

Translated fromChinese

本发明实施例提供了一种填槽方法和装置，所述方法包括：加载模板引擎；所述模板引擎包括模板信息文件和填槽信息文件；针对获取到的语音请求，生成匹配的语音识别信息；确定所述模板信息文件中与所述语音识别信息匹配的目标模板；若是，则确定在所述填槽信息文件与所述语音识别信息对应的槽位信息；采用所述待填充槽以及所述槽位信息生成填槽结果。本发明实施例可以实现在模板引擎中添加填槽功能，能够在与语音识别信息匹配的目标模板存在待填充槽时，在填槽信息文件中检索对应的槽位信息并生成相应的填槽结果。

Embodiments of the present invention provide a slot filling method and device. The method includes: loading a template engine; the template engine includes a template information file and a slot filling information file; and generating matching voice recognition information for an acquired voice request Determine the target template matched with the speech recognition information in the template information file; if so, then determine the slot information corresponding to the speech recognition information in the slot filling information file; adopt the slot to be filled and all The slot information is used to generate the slot filling result. The embodiment of the present invention can realize adding a slot filling function to the template engine, and can retrieve the corresponding slot information in the slot filling information file and generate the corresponding slot filling result when there is a slot to be filled in the target template matching the speech recognition information. .

Description

Groove filling method and device

Technical Field

The invention relates to the technical field of computers, in particular to a groove filling method and a groove filling device.

Background

Natural Language Processing (NLP) is a technology for communicating with a computer using a Natural Language, and because the key to Processing the Natural Language is to "understand" the Natural Language, the Natural Language Processing is also called Natural Language Understanding (NLU).

In the NLU scheme, a language model is generally provided for natural language understanding, and a template parser is provided for parsing a common sentence pattern, so as to make up for the lack of accuracy of the model. The template supports the expression of some specific sentence patterns, but the dynamic keywords corresponding to some sentence patterns are dynamically updated, taking a vehicle-mounted service scene as an example, a vehicle-mounted system is provided with a third party application and a third party applet, and the template engine is adopted to support the opening and closing of the third party application and the third party applet.

Because the business generalization statement is more, the white list sentence pattern is difficult to support the identification of all third-party applications and third-party applets.

Disclosure of Invention

In view of the above, embodiments of the present invention are proposed in order to provide a caulking method and a corresponding caulking apparatus that overcome or at least partially solve the above problems.

In order to solve the above problem, an embodiment of the present invention discloses a slot filling method, including:

loading a template engine; the template engine comprises a template information file and a slot filling information file;

generating matched voice recognition information aiming at the acquired voice request;

determining a target template matched with the voice recognition information in the template information file;

judging whether the target template is provided with a groove to be filled;

if so, determining the slot position information corresponding to the voice recognition information in the slot filling information file;

and generating a slot filling result by adopting the slot to be filled and the slot position information.

Optionally, the method further comprises:

recognizing intention information and field information matched with the voice recognition information by adopting the target template;

and generating instruction information by adopting the slot position information, the intention information and the field information.

Optionally, the template information file records a plurality of preset templates; the step of determining a target template matching the speech recognition information in the template information file includes:

extracting key information in the voice recognition information;

and determining a target template matched with the key information in the preset templates.

Optionally, the step of determining a target template matching the key information in the preset templates includes:

determining at least one keyword set in a preset template; the keyword set comprises one or more keywords;

and determining a preset template matched with the key information as a target template.

Optionally, the step of determining slot position information corresponding to the voice recognition information in the slot filling information file includes:

determining a target keyword set corresponding to the slot to be filled;

determining target keywords matched with the key information in the target keyword set;

and determining slot position information corresponding to the slot filling information file and the target keyword.

Optionally, the method further comprises:

and when detecting that the database corresponding to the slot filling information file changes, updating the slot filling information file according to the current content of the database.

Optionally, the slot information includes: slot name, normalized value, and slot value type.

The embodiment of the invention also discloses a slot filling device, which comprises:

the loading module is used for loading the template engine; the template engine comprises a template information file and a slot filling information file;

the voice recognition module is used for generating matched voice recognition information according to the acquired voice request;

the target template determining module is used for determining a target template matched with the voice recognition information in the template information file;

the groove to be filled judging module is used for judging whether the target template is provided with a groove to be filled;

the slot position information determining module is used for determining slot position information corresponding to the voice recognition information in the slot filling information file if the target template is provided with a slot to be filled;

and the slot filling result module is used for generating a slot filling result by adopting the slot to be filled and the slot position information.

The embodiment of the invention also discloses a vehicle which comprises a processor, a memory and a computer program which is stored on the memory and can run on the processor, wherein the computer program realizes the slot filling method when being executed by the processor.

The embodiment of the invention also discloses a computer readable storage medium, wherein a computer program is stored on the computer readable storage medium, and when the computer program is executed by a processor, the slot filling method is realized.

The embodiment of the invention has the following advantages:

obtaining a template information file and a slot filling information file by loading a template engine, automatically carrying out voice recognition on an obtained voice request to obtain matched voice recognition information, determining a target template matched in the template information file according to the voice recognition information, retrieving slot position information matched with the voice recognition information in the slot filling information file when the target template is determined to have a slot to be filled, filling the slot position information into the slot to be filled, generating a slot filling result, thereby realizing matching between the template information file in the template engine and the voice recognition information, determining a corresponding target template, retrieving the slot filling information file according to the voice recognition information when the target template has the slot to be filled, retrieving the corresponding slot position information, filling the slot to be filled in the target template with the slot position information, generating the slot filling result, and further realizing the function of adding the slot filling in the template engine, when a target template matched with the voice recognition information has a slot to be filled, searching corresponding slot position information in a slot filling information file and generating a corresponding slot filling result, wherein the slot filling information file is a dynamically updated file, and when the voice recognition information relates to dynamic content, the slot filling information file can be used for filling the voice recognition information by dynamically updating the slot filling information file.

Drawings

FIG. 1 is a flowchart illustrating steps of a first embodiment of a slot filling method according to the present invention;

FIG. 2 is a flowchart illustrating steps of a second embodiment of a slot filling method according to the present invention;

FIG. 3 is a flowchart illustrating an exemplary method of filling a slot according to the present invention;

fig. 4 is a block diagram of a slot filling apparatus according to an embodiment of the present invention.

Detailed Description

In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.

One of the core concepts of the embodiments of the present invention is to set a slot to be filled in a template recorded in a template information file of a template engine and dynamically update the slot filling information file in the template engine, when obtaining voice recognition information matched with a voice request, and when determining that a target template corresponding to the voice recognition information exists in the slot to be filled according to the template information file in the template engine, retrieve slot filling information corresponding to the voice recognition information in the slot filling information file, and fill the slot to be filled with the slot filling information to generate a slot filling result, thereby implementing slot filling for dynamically updated contents related to the voice recognition information by using the template.

Referring to fig. 1, a flowchart illustrating steps of a first embodiment of a slot filling method according to the present invention is shown, which may specifically include the following steps:

step 101, loading a template engine; the template engine comprises a template information file and a slot filling information file;

the embodiment of the invention can be applied to the server, and the server can be connected with a vehicle-mounted system in the vehicle to realize the connection communication between the server and the vehicle. The server can be pre-deployed with a template engine, and the template engine is provided with a template information file and a slot filling information file. The template information file is used for recording a preset template, and the slot filling information file is used for recording preset slot position information.

By loading the template engine, the template information file and the slot filling information file can be read.

The slot position information is dynamically updated information, a service platform can be deployed in the server, and the slot position information is dynamically updated through the service platform.

102, generating matched voice recognition information aiming at the acquired voice request;

the vehicle-mounted system can receive a voice request of a user and send the voice request to the server. The server can perform ASR (Automatic Speech Recognition) for the acquired Speech request, and generate Speech Recognition information matching the Speech request.

In particular implementations, the speech recognition information may be textual information.

103, determining a target template matched with the voice recognition information in the template information file;

and performing natural language understanding on the voice recognition information, judging whether the voice recognition information is matched with any preset template in a target information file, and if so, determining a target template matched with the voice recognition information in the template information file.

One or more preset templates can be recorded in the template information file, the preset templates can have certain template grammar, the template grammar comprises at least two parts, one part is of a sentence pattern structure and is used for matching with the voice recognition information, and the other part is a semantic result. For example: the template grammar can be [ K: open ] [ K: third _ applet _ name ] [ 'gui @ gui _ applet _ open' ], wherein the part before the colon in the preset template is a sentence structure, the sentence structure can contain a keyword set, K represents a keyword (key), K: open represents an open keyword set, and represents a series of synonym sets of "open", such as open, and the like. Three (third) _ applet (name) represents a third party applet name keyword set and represents a series of applet names on the vehicle-mounted system; the part after the colon is the semantic result. When voice recognition information matched with K: open K: third _ applet _ name exists, outputting the semantic result of the voice recognition information as GUI _ applet _ open intention in the field of GUI (Graphical User Interface), namely opening program intention in the Graphical User Interface.

When natural language understanding is carried out on voice recognition information and the voice recognition information is judged to also contain an opening keyword and a third-party small program name keyword, a preset template matched with the voice recognition information is determined as a target template.

104, judging whether the target template is provided with a groove to be filled;

in the preset template, there may be a case where some slots to be filled are provided and some slots to be filled are not provided, so that it is necessary to identify whether slots to be filled exist in the target template. Specifically, when a keyword set exists in a semantic result part in the target template, the slot to be filled is determined to exist.

For example: the target template is "[ K: open ] [ K: third _ applet _ name ] [ 'gui @ gui _ applet _ open K: third _ applet _ name' ]" to determine that the slot K: third _ applet _ name to be filled exists.

Step 105, if yes, determining slot position information corresponding to the voice recognition information in the slot filling information file;

and if the target template has a slot to be filled, determining slot position information corresponding to the voice recognition information in the slot filling information file. Specifically, when the target template has a slot to be filled, the slot filling information file can be retrieved according to the voice recognition information, and the slot position information corresponding to the voice recognition information is determined.

And 106, generating a slot filling result by adopting the slot to be filled and the slot position information.

And filling the slot to be filled by adopting the slot position information to generate a slot filling result.

Taking the voice recognition information as an example of opening the a program, the slot to be filled corresponds to a program name, and slot information (including but not limited to a label name and an alias) corresponding to a plurality of program names is recorded in the slot filling information file, when the target template has the slot to be filled, the slot information corresponding to the a program is retrieved from the slot filling information file, and the obtained slot filling information is filled into the slot to be filled, so as to obtain a slot filling result.

In a practical application, the slot to be filled has identification information, and a slot filling result is generated based on the identification information and the slot filling information.

In the embodiment of the invention, a template information file and a slot filling information file are obtained by loading a template engine, the obtained voice request is automatically subjected to voice recognition to obtain matched voice recognition information, a target template matched in the template information file is determined according to the voice recognition information, when the target template is determined to have a slot to be filled, slot position information matched with the voice recognition information in the slot filling information file is retrieved, the slot position information is filled into the slot to be filled, a slot filling result is generated, so that the matching between the template information file in the template engine and the voice recognition information is realized, a corresponding target template is determined, when the target template has the slot to be filled, the slot filling information file is retrieved according to the voice recognition information, the corresponding slot position information is retrieved, the slot to be filled in the target template is filled with the slot position information, the slot filling result is generated, and the function of adding the slot filling in the template engine is realized, when a target template matched with the voice recognition information has a slot to be filled, searching corresponding slot position information in a slot filling information file and generating a corresponding slot filling result, wherein the slot filling information file is a dynamically updated file, and when the voice recognition information relates to dynamic content, the slot filling information file can be used for filling the voice recognition information by dynamically updating the slot filling information file.

Referring to fig. 2, a flowchart illustrating steps of a second embodiment of the groove filling method of the present invention is shown, which may specifically include the following steps:

step 201, loading a template engine; the template engine comprises a template information file and a slot filling information file;

step 202, generating matched voice recognition information aiming at the acquired voice request;

step 203, determining a target template matched with the voice recognition information in the template information file;

in an optional embodiment of the present invention, the template information file records a plurality of preset templates;step 203 may comprise:

substep S11, extracting key information in the speech recognition information;

the key information is information of a specified type, and taking the action type as an example, the key information can be opened, closed and the like; taking the type of the program as an example, the key information may be a standard name of the program, an alias of the program, etc., and the program may include an application program and/or an applet.

Specifically, the key information in the speech recognition information can be extracted through the trained model.

And a substep S12, determining a target template matching the key information in the preset templates.

The template information file records a plurality of different preset templates, wherein the preset template matched with the current voice recognition information is determined as a target template.

In an alternative embodiment of the present invention, the sub-step S12 may include:

substep S121, determining at least one keyword set in a preset template; the keyword set comprises one or more keywords;

and a substep S122, determining the preset template matched with the key words and the key information as a target template.

And when the key information corresponds to the key word set, determining that the template matched with the key information in the key word set is the target template.

Taking the template [ K: open ] [ K: third _ applet _ name ] [ 'gui @ gui _ applet _ open K: third _ applet _ name' ], two keyword sets are included: [ K: open ] and [ K: third applet name ], [ K: open ] contains one or more keywords synonymous with "open", and [ K: third applet name ] contains one or more program names as keywords. And when the voice recognition information is 'open XY carriage', determining that the keyword information is 'open' and 'XY carriage' are respectively matched with [ K: open ] and [ K: third _ applet _ name ] in the template, and determining that the template is the target template.

In an example, when key information of the voice recognition information matches a plurality of preset templates, a matching degree of the key information and the matching preset templates may be determined, and one with the largest matching degree may be determined as the target template.

Step 204, judging whether the target template is provided with a groove to be filled;

step 205, if yes, determining slot position information corresponding to the voice recognition information in the slot filling information file;

in an optional embodiment of the present invention, the step of determining slot position information corresponding to the speech recognition information in the slot filling information file includes:

substep S21, determining a target keyword set corresponding to the slot to be filled;

the slot to be filled may have fields consistent with the target key, such as: in the template [ K: open ] [ K: third _ applet _ name ] [ 'gui @ gui _ applet _ open K: third _ applet _ name' ], the target keyword set corresponding to the slot K: third _ applet _ name to be filled is [ K: third _ applet _ name ].

Substep S22, determining a target keyword in the target keyword set, which matches the key information;

and determining one matched with the key information as the target key word in one or more key words contained in the target key word set.

And a substep S23 of determining slot position information corresponding to the target keyword in the slot filling information file.

The slot filling information file records information about the content to be extracted from the voice recognition information. For example: the slot filling information file may record slot position information related to the program, and when the target keyword is a program name, the slot position information corresponding to the program name is determined in the slot filling information file.

In another example, the slot filling information file may be recorded with slot information related to music, and when the target keyword is a song name, the slot information corresponding to the song name is determined in the slot filling information file.

Step 206, generating a slot filling result by using the slot to be filled and the slot position information;

in an optional embodiment of the present invention, the slot information includes: slot name, normalized value, and slot value type.

In the embodiment of the present invention, the slot filling result can be divided into three levels of information: the first level is the keyword name, i.e. the keyword name is extracted as the slot position. The second level is the specific entry in the keyword. The third level is the specific slot position information, and the slot position information is embodied in a dictionary form after being loaded. The front part of the colon of each line is used as a key word, and the rear part of the colon is used as a value corresponding to the key word. In the present embodiment, the slot information includes a slot name (name), a normalized value (value), and a slot value type (value _ type). It can be understood that in practical application, the content of the slot position information can be increased or decreased, and other custom information can be added to the slot position information according to requirements.

In an example, the voice recognition information is "open XY car keeping", the matched target template is [ K: open ] [ K: third _ applet _ name ] [ 'gui @ gui _ applet _ open K: third _ applet _ name' ], the obtained slot to be filled is K: third _ applet _ name, the obtained target keyword is "XY car keeping", the obtained matched slot information is name: applet _ name, value: XY, value _ type: string, and the obtained slot filling result is: k, third _ applet _ name, XY carriage (entry), name, applet _ name, value, XY, value type, string.

Step 207, recognizing intention information and field information matched with the voice recognition information by using the target template;

and after the target template is determined, adopting the target template to identify intention information and field information matched with the voice identification information. For example: the target template may be [ K: open ] [ K: third _ applet _ name ] [ 'gui @ gui _ applet _ open' ], the intention information matched with the voice recognition information by using template target recognition is gui _ applet _ open, i.e., the graphic user interface opening intention, and the domain information is gui domain, i.e., the graphic user interface domain.

And 208, generating instruction information by adopting the slot position information, the intention information and the field information.

And generating corresponding instruction information by adopting the slot position information, the intention information and the field information obtained based on the target template. The vehicle-mounted terminal and/or the server can respond to the instruction information and execute the operation matched with the slot position information, the intention information and the field information.

The example is that the voice recognition information is "open a program", and the in-vehicle terminal is provided with a display component for displaying a graphical user interface. And when the obtained slot position information and the intention information corresponding to the program A are that the program is opened in the graphical user interface and the field information is that the graphical user interface is in the field, generating corresponding instruction information, and enabling the vehicle-mounted terminal to respond to the instruction information and open the program A in the graphical user interface displayed by the display component.

It will be appreciated that the intent information and domain information may also be information related to a non-graphical user interface, such as: when the voice recognition information is "search XXX song", the domain information may be a music domain, the intention information may be a search song, the slot information may be information corresponding to "XXX", and "XXX" may be a person name, an album name, a song name, a music company name, or the like.

In an optional embodiment of the invention, the method further comprises:

and when detecting that the database corresponding to the slot filling information file changes, updating the slot filling information file according to the current content of the database.

The template engine can detect the database corresponding to the slot filling information file according to the designated frequency, and when the database corresponding to the slot filling information file is detected to be changed, the slot filling information file is updated according to the current content of the database.

In one example, the template engine listens for database update events and, when a database update is received, updates the slot filling information file according to the current contents of the database.

The slot position information may correspond to a program, the database may include a program for putting up, and when a new program is put up or a program is put down, the slot filling information file is updated according to the current program for putting up of the database.

The slot position information may correspond to songs, the database may include songs that are on shelf, and when a new song is on shelf or a song is off shelf, the slot filling information file is updated according to the current song on shelf in the database.

It can be understood that the slot filling information file may also correspond to other contents, and when the corresponding contents change, the slot filling information file is updated dynamically, and the content included in the slot position information is not limited in the embodiment of the present invention.

In the embodiment of the invention, a template information file and a slot filling information file are obtained by loading a template engine, the obtained voice request is automatically subjected to voice recognition to obtain matched voice recognition information, the key information in the voice recognition information is matched with a keyword set contained in a preset template recorded in the template information file, the preset template with the keyword set matched with the key information determines a target template, when the target template is determined to have a slot to be filled, slot position information matched with the voice recognition information in the slot filling information file is searched, the slot position information is filled into the slot to be filled, a slot filling result is generated, so that the matching between the template information file in the template engine and the voice recognition information is realized, a corresponding target template is determined, and when the target template has the slot to be filled, the slot filling information file is searched according to the voice recognition information, the corresponding slot position information is retrieved, the slot to be filled in the slot target template is filled with the slot position information, a slot filling result is generated, therefore, the slot filling function is added in the template engine, when the slot to be filled exists in the target template matched with the voice recognition information, the corresponding slot position information is retrieved from the slot filling information file, the corresponding slot filling result is generated, the slot filling information file is a dynamically updated file, and when the voice recognition information relates to dynamic content, the slot filling information file can be used for filling the voice recognition information by dynamically updating the slot filling information file. Further, the field information and the intention information of the voice recognition information are recognized through the target template, the slot position information, the field information and the intention information are adopted to obtain the instruction information, and the operation corresponding to the voice request is executed through responding to the instruction information.

The following further illustrates an embodiment of the present invention by way of an example. Referring to fig. 3, a flowchart of an exemplary slot filling method of the present invention is shown, and the exemplary slot filling method may include the following steps:

step 301, pre-loading a template engine, including a template information file and a slot filling information file of a local andskill management platform 310, wherein the skill management platform is equivalent to a dynamic data management platform and is used for maintaining dynamic data. The skill management platform can perform hot update on the template information file.

Atstep 302, the NLU service is started and the server performs NLU processing on the query (i.e., voice request).

Step 303: and judging whether the template accords with a predefined template or not, and if so, judging whether the slot lifting is required or not. If the groove lifting is not needed, namely, the groove lifting key words are not in the template. Then the semantic result of the template matching is returned and the template matching is finished. If a lifting of the slot is required,step 304 is performed.

Step 304: and extracting the slot position, correspondingly retrieving the preloaded slot filling information, obtaining a slot filling result, returning, and finishing template matching.

The third-party application &applet data 320 is dynamically updated data, maintained by the skill management platform, and is synchronized to the skill management platform when the data is changed, and then the skill platform synchronously updates the data to the online NLU service, and the NLU service performs a hot update (template hot update) to update the template information file and the slot filling information file.

It should be noted that, for simplicity of description, the method embodiments are described as a series of acts or combination of acts, but those skilled in the art will recognize that the present invention is not limited by the illustrated order of acts, as some steps may occur in other orders or concurrently in accordance with the embodiments of the present invention. Further, those skilled in the art will appreciate that the embodiments described in the specification are presently preferred and that no particular act is required to implement the invention.

Referring to fig. 4, a block diagram of a structure of an embodiment of a slot filling apparatus of the present invention is shown, which may specifically include the following modules:

aloading module 401, configured to load a template engine; the template engine comprises a template information file and a slot filling information file;

avoice recognition module 402, configured to generate matching voice recognition information for the obtained voice request;

a targettemplate determining module 403, configured to determine a target template in the template information file, where the target template matches the speech recognition information;

a slot to be filled judgingmodule 404, configured to judge whether a slot to be filled is set in the target template;

a slot positioninformation determining module 405, configured to determine slot position information corresponding to the speech recognition information in the slot filling information file if a slot to be filled is set in the target template;

and a slot fillingresult module 406, configured to generate a slot filling result by using the slot to be filled and the slot position information.

In an optional embodiment of the invention, the apparatus further comprises:

the domain intention recognition module is used for recognizing intention information and domain information matched with the voice recognition information by adopting the target template;

and the instruction information generating module is used for generating instruction information by adopting the slot position information, the intention information and the field information.

In an optional embodiment of the present invention, the template information file records a plurality of preset templates; the targettemplate determination module 403 includes:

the key information extraction submodule is used for extracting key information in the voice recognition information;

and the target template matching submodule is used for determining a target template which is matched with the key information in the preset templates.

In an optional embodiment of the invention, the target template matching sub-module comprises:

the keyword set unit is used for determining at least one keyword set in a preset template; the keyword set comprises one or more keywords;

and the target template unit is used for determining the preset template matched with the key information as the target template.

In an optional embodiment of the present invention, the slotinformation determining module 405 includes:

the target keyword set submodule is used for determining a target keyword set corresponding to the slot to be filled;

the target keyword submodule is used for determining target keywords matched with the key information in the target keyword set;

and the slot position information submodule is used for determining slot position information corresponding to the slot filling information file and the target keyword.

In an optional embodiment of the invention, the apparatus further comprises:

and the updating module is used for updating the slot filling information file according to the current content of the database when detecting that the database corresponding to the slot filling information file changes.

In an optional embodiment of the present invention, the slot information includes: slot name, normalized value, and slot value type.

For the device embodiment, since it is basically similar to the method embodiment, the description is simple, and for the relevant points, refer to the partial description of the method embodiment.

The embodiment of the invention also discloses a vehicle which comprises a processor, a memory and a computer program which is stored on the memory and can run on the processor, wherein the computer program realizes the slot filling method when being executed by the processor.

The embodiment of the invention also discloses a computer readable storage medium, wherein a computer program is stored on the computer readable storage medium, and when the computer program is executed by a processor, the slot filling method is realized.

The embodiments in the present specification are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other.

As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, apparatus, or computer program product. Accordingly, embodiments of the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

Embodiments of the present invention are described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing terminal to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing terminal to cause a series of operational steps to be performed on the computer or other programmable terminal to produce a computer implemented process such that the instructions which execute on the computer or other programmable terminal provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

While preferred embodiments of the present invention have been described, additional variations and modifications of these embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the embodiments of the invention.

Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or terminal. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or terminal that comprises the element.

The groove filling method and device provided by the invention are described in detail, and the principle and the implementation mode of the invention are explained by applying specific examples, and the description of the examples is only used for helping to understand the method and the core idea of the invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.

Claims (10)

Translated fromChinese

1.一种填槽方法，其特征在于，包括：1. a groove filling method, is characterized in that, comprises:加载模板引擎；所述模板引擎包括模板信息文件和填槽信息文件；Loading a template engine; the template engine includes a template information file and a slot filling information file;针对获取到的语音请求，生成匹配的语音识别信息；For the acquired voice request, generate matching voice recognition information;确定所述模板信息文件中与所述语音识别信息匹配的目标模板；Determine the target template matched with the speech recognition information in the template information file;判断所述目标模板是否设置有待填充槽；Determine whether the target template is provided with a slot to be filled;若是，则确定在所述填槽信息文件与所述语音识别信息对应的槽位信息；If yes, then determine the slot information corresponding to the voice recognition information in the slot filling information file;采用所述待填充槽以及所述槽位信息生成填槽结果。A slot filling result is generated by using the slot to be filled and the slot position information.2.根据权利要求1所述的方法，其特征在于，还包括：2. The method of claim 1, further comprising:采用所述目标模板识别与所述语音识别信息匹配的意图信息以及领域信息；Use the target template to identify intent information and domain information that match the speech recognition information;采用所述槽位信息、所述意图信息以及所述领域信息，生成指令信息。Using the slot information, the intent information, and the domain information, instruction information is generated.3.根据权利要求1或2所述的方法，其特征在于，所述模板信息文件记录有多个预设模板；所述确定所述模板信息文件中与所述语音识别信息匹配的目标模板的步骤，包括：3. The method according to claim 1 or 2, wherein the template information file is recorded with a plurality of preset templates; the determination of the target template matching the speech recognition information in the template information file steps, including:提取所述语音识别信息中的关键信息；extracting key information in the speech recognition information;确定所述预设模板中与所述关键信息匹配的目标模板。A target template in the preset template that matches the key information is determined.4.根据权利要求3所述的方法，其特征在于，所述确定所述预设模板中与所述关键信息匹配的目标模板的步骤，包括：4. The method according to claim 3, wherein the step of determining the target template matching the key information in the preset template comprises:确定预设模板中的至少一个关键词集合；所述关键词集合包含一个或多个关键词；Determine at least one keyword set in the preset template; the keyword set contains one or more keywords;确定所述关键词与所述关键信息匹配的预设模板为目标模板。A preset template that matches the keyword with the key information is determined as a target template.5.根据权利要求4所述的方法，其特征在于，所述确定在所述填槽信息文件与所述语音识别信息对应的槽位信息的步骤，包括：5. The method according to claim 4, wherein the step of determining the slot information corresponding to the voice recognition information in the slot filling information file comprises:确定与所述待填充槽对应的目标关键词集合；determining a target keyword set corresponding to the slot to be filled;确定所述目标关键词集合中与所述关键信息匹配的目标关键词；determining the target keyword matching the key information in the target keyword set;确定所述填槽信息文件与所述目标关键词对应的槽位信息。Determine the slot information corresponding to the target keyword in the slot filling information file.6.根据权利要求1或2所述的方法，其特征在于，所述方法还包括：6. The method according to claim 1 or 2, wherein the method further comprises:在检测到与所述填槽信息文件对应的数据库发生变化时，按照所述数据库当前的内容更新所述填槽信息文件。When it is detected that the database corresponding to the slot filling information file changes, the slot filling information file is updated according to the current content of the database.7.根据权利要求1或2所述的方法，其特征在于，所述槽位信息包括：槽位名、归一化值以及槽值类型。7. The method according to claim 1 or 2, wherein the slot information comprises: a slot name, a normalized value, and a slot value type.8.一种填槽装置，其特征在于，包括：8. A groove filling device, characterized in that, comprising:加载模块，用于加载模板引擎；所述模板引擎包括模板信息文件和填槽信息文件；a loading module for loading a template engine; the template engine includes a template information file and a slot filling information file;语音识别模块，用于针对获取到的语音请求，生成匹配的语音识别信息；The speech recognition module is used to generate matching speech recognition information for the acquired speech request;目标模板确定模块，用于确定所述模板信息文件中与所述语音识别信息匹配的目标模板；a target template determination module, for determining a target template matching the speech recognition information in the template information file;待填充槽判断模块，用于判断所述目标模板是否设置有待填充槽；a slot to be filled judging module, used to determine whether the target template is provided with a slot to be filled;槽位信息确定模块，用于若所述目标模板设置有待填充槽，则确定在所述填槽信息文件与所述语音识别信息对应的槽位信息；a slot information determination module, configured to determine the slot information corresponding to the voice recognition information in the slot filling information file if the target template is provided with a slot to be filled;填槽结果模块，用于采用所述待填充槽以及所述槽位信息生成填槽结果。A slot filling result module, configured to generate a slot filling result by using the to-be-filled slot and the slot position information.9.一种车辆，其特征在于，包括处理器、存储器及存储在所述存储器上并能够在所述处理器上运行的计算机程序，所述计算机程序被所述处理器执行时实现如权利要求1至7中任一项所述的填槽方法。9. A vehicle, characterized in that it comprises a processor, a memory and a computer program stored on the memory and capable of running on the processor, the computer program being executed by the processor to achieve the invention as claimed in the claims The groove filling method described in any one of 1 to 7.10.一种计算机可读存储介质，其特征在于，所述计算机可读存储介质上存储计算机程序，所述计算机程序被处理器执行时实现如权利要求1至7中任一项所述的填槽方法。10. A computer-readable storage medium, characterized in that, a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the filling method according to any one of claims 1 to 7 is realized. slot method.

Images