CN112259102A - Retail scene voice interaction optimization method based on knowledge graph - Google Patents
Retail scene voice interaction optimization method based on knowledge graph Download PDFInfo
- Publication number
- CN112259102A CN112259102A CN202011179681.4A CN202011179681A CN112259102A CN 112259102 A CN112259102 A CN 112259102A CN 202011179681 A CN202011179681 A CN 202011179681A CN 112259102 A CN112259102 A CN 112259102A
- Authority
- CN
- China
- Prior art keywords
- voice
- recognition
- brand
- scene
- knowledge graph
- 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
- 238000000034 method Methods 0.000 title claims abstract description 34
- 230000003993 interaction Effects 0.000 title claims abstract description 14
- 238000005457 optimization Methods 0.000 title claims abstract description 13
- 238000012549 training Methods 0.000 claims abstract description 14
- 230000002452 interceptive effect Effects 0.000 claims abstract description 13
- 230000000694 effects Effects 0.000 claims abstract description 6
- 238000012360 testing method Methods 0.000 claims abstract description 6
- 238000010801 machine learning Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 230000003442 weekly effect Effects 0.000 claims 1
- 238000000605 extraction Methods 0.000 abstract description 5
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/005—Language recognition
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/02—Feature extraction for speech recognition; Selection of recognition unit
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/06—Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice
- G10L15/063—Training
Abstract
The invention discloses a retail scene voice interaction optimization method based on a knowledge graph, which comprises the following steps: s1, inputting voice, and performing language recognition by the system through the processor; s2, extracting labels and association relations from brand commodity information, and constructing a brand knowledge graph; s3, voice interactive shopping searching based on the online knowledge map; s4, analyzing the recorded voice data, and adding homophones and synonyms; s5, applying the new data training model to the business, and greatly improving the use experience in the voice interaction shopping scene. The brand knowledge map is established by receiving brand professional words and learning different regional pronunciations and synonyms in the whole country, and the voice recognition effect can be greatly improved by combining with voice recognition and NLP. According to the test, the recognition of the special brand words reaches 92%, the text intention extraction accuracy reaches 96%, and the relation understanding accuracy in the scene reaches 90%. Meanwhile, scene word search is optimized, and commodity display and transaction are greatly promoted in voice interactive shopping.
Description
Technical Field
The invention relates to the technical field of voice interaction, in particular to a retail scene voice interaction optimization method based on a knowledge graph.
Background
The voice is the first attribute among three basic attributes of the shape, the sound and the meaning of the language, the human language is formed in the form of voice firstly, and the voice has a language without characters in the world but does not have the language without voice, and plays a decisive supporting role in the language;
however, the retail scene voice interaction optimization method based on the knowledge graph in the market at present only is suitable for individual customization and use by directly carrying out personalized training on the model, the difficulty in use in the vertical industry is high, the model is difficult to train, each brand has own exclusive knowledge content and is difficult to train, and retail commodities are updated quickly.
Disclosure of Invention
The invention provides a retail scene voice interaction optimization method based on a knowledge graph, which can effectively solve the problems that the retail scene voice interaction optimization method based on the knowledge graph in the current market is provided in the background technology, the model is directly trained in an individualized way and is only suitable for individual customization, the use difficulty in the vertical industry is high, the model is difficult to train, each brand has own exclusive knowledge content and is not easy to train, and retail commodities are updated quickly.
In order to achieve the purpose, the invention provides the following technical scheme: a retail scene voice interaction optimization method based on a knowledge graph comprises the following steps:
s1, inputting voice, and performing language recognition by the system through the processor;
s2, extracting labels and association relations from brand commodity information, and constructing a brand knowledge graph;
s3, voice interactive shopping searching based on the online knowledge map;
s4, analyzing the recorded voice data, and adding homophones and synonyms;
and S5, applying the new data training model to the business.
According to the technical scheme, learning is carried out through reading linguistic data in the S1, a word stock of the user is established, and the dialect identification problem is solved;
according to the agreement of users, the special calling methods and different accents of commodities in different regions in a wide area are collected anonymously, in a voice shopping guide scene, default voices are related to services, and corpora are collected through machine learning and supervised learning to build a word stock in a vertical industry.
According to the technical scheme, the brand knowledge map is established in the S2 by receiving brand professional words and learning pronunciations and synonyms of different regions in the country, and the voice recognition effect can be greatly improved by combining with voice recognition and NLP.
According to the technical scheme, the accuracy of brand special word recognition, special word meaning analysis and scene relation analysis in the step S2 are matched with each other.
According to the technical scheme, data in the voice interactive shopping search in the S3 are updated every week, the data are stored in the server and are received and transmitted through the system;
the system performs voice recognition and input through network and personal input.
According to the technical scheme, the homophones and the synonyms in the S4 are listed according to original words, are divided into homophones, harmonious characters and harmonious characters, and comprise numbers, Arabic letters and a few voices.
According to the technical scheme, the step S5 is applied to the service, the voice data is transmitted to the voice system, the data of the voice data is stored and applied, and a loop is formed with the step S1.
According to the above technical solution, the complete specific person isolated word speech recognition system in S1 generally includes speech input, speech signal preprocessing, feature extraction, training and recognition;
the process of speech recognition can be regarded as the process of pattern matching, the process of pattern matching refers to the process of making unknown pattern and a certain model in the model base obtain the best matching according to certain criteria, the reference template that needs to be used in the pattern matching is obtained through template training, in the training stage, after certain processing is carried out on the characteristic parameters, a model is established for each entry and is stored as the template base, in the recognition stage, speech signals pass through the same channel to obtain speech characteristic parameters, a test template is generated and is matched with the reference template, the reference template with the highest matching score is used as the recognition result, and meanwhile, the accuracy of recognition can be improved with the help of a priori knowledge.
Compared with the prior art, the invention has the beneficial effects that: the voice interactive shopping system is scientific and reasonable in structure, safe and convenient to use, and greatly improves the use experience in the voice interactive shopping scene. The brand knowledge map is established by receiving brand professional words and learning different regional pronunciations and synonyms in the whole country, and the voice recognition effect can be greatly improved by combining with voice recognition and NLP. According to the test, the recognition of the special brand words reaches 92%, the text intention extraction accuracy reaches 96%, and the relation understanding accuracy in the scene reaches 90%. Meanwhile, scene word search is optimized, and commodity display and transaction are greatly promoted in voice interactive shopping.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention.
In the drawings:
FIG. 1 is a schematic flow chart diagram illustrating the steps of the optimization method of the present invention;
FIG. 2 is a data comparison diagram of the present invention.
Detailed Description
The preferred embodiments of the present invention will be described in conjunction with the accompanying drawings, and it will be understood that they are described herein for the purpose of illustration and explanation and not limitation.
Example (b): as shown in fig. 1-2, the invention provides a technical solution, a retail scene voice interaction optimization method based on knowledge graph, comprising the following steps:
s1, inputting voice, and performing language recognition by the system through the processor;
s2, extracting labels and association relations from brand commodity information, and constructing a brand knowledge graph;
s3, voice interactive shopping searching based on the online knowledge map;
s4, analyzing the recorded voice data, and adding homophones and synonyms;
and S5, applying the new data training model to the business.
According to the technical scheme, learning is carried out through reading linguistic data in S1, a word stock of the user is established, and the dialect identification problem is solved;
according to the agreement of users, the special calling methods and different accents of commodities in different regions in a wide area are collected anonymously, in a voice shopping guide scene, default voices are related to services, and corpora are collected through machine learning and supervised learning to build a word stock in a vertical industry.
According to the technical scheme, the brand knowledge map is established in the S2 by receiving and recording brand professional words and learning pronunciations and synonyms of different regions in the country, and the voice recognition effect can be greatly improved by combining with voice recognition and NLP.
According to the technical scheme, the accuracy of brand special word recognition, special word meaning analysis and scene relation analysis in the S2 are matched with each other.
According to the technical scheme, data in voice interactive shopping search in the S3 are updated every week, the data are stored in the server and are received and transmitted through the system;
the system performs voice recognition and input through network and personal input.
According to the technical scheme, homophones and synonyms in the S4 are listed according to original words, are divided into homophones, harmonious characters and harmonious characters, and comprise numbers, Arabic letters and a few voices.
According to the technical scheme, the S5 is applied to the service, the voice data is transmitted to the voice system, the data of the voice data is stored and applied, and a loop is formed with the step S1.
According to the technical scheme, the complete specific person isolated word voice recognition system in the S1 generally comprises voice input, voice signal preprocessing, feature extraction, training and recognition;
the process of speech recognition can be regarded as the process of pattern matching, the process of pattern matching refers to the process of making unknown pattern and a certain model in the model base obtain the best matching according to certain criteria, the reference template that needs to be used in the pattern matching is obtained through template training, in the training stage, after certain processing is carried out on the characteristic parameters, a model is established for each entry and is stored as the template base, in the recognition stage, speech signals pass through the same channel to obtain speech characteristic parameters, a test template is generated and is matched with the reference template, the reference template with the highest matching score is used as the recognition result, and meanwhile, the accuracy of recognition can be improved with the help of a priori knowledge.
Compared with the prior art, the invention has the beneficial effects that: the voice interactive shopping system is scientific and reasonable in structure, safe and convenient to use, and greatly improves the use experience in the voice interactive shopping scene. The brand knowledge map is established by receiving brand professional words and learning different regional pronunciations and synonyms in the whole country, and the voice recognition effect can be greatly improved by combining with voice recognition and NLP. According to the test, the recognition of the special brand words reaches 92%, the text intention extraction accuracy reaches 96%, and the relation understanding accuracy in the scene reaches 90%. Meanwhile, scene word search is optimized, and commodity display and transaction are greatly promoted in voice interactive shopping.
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that changes may be made in the embodiments and/or equivalents thereof without departing from the spirit and scope of the invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (8)
1. A retail scene voice interaction optimization method based on a knowledge graph is characterized by comprising the following steps: the method comprises the following steps:
s1, inputting voice, and performing language recognition by the system through the processor;
s2, extracting labels and association relations from brand commodity information, and constructing a brand knowledge graph;
s3, voice interactive shopping searching based on the online knowledge map;
s4, analyzing the recorded voice data, and adding homophones and synonyms;
and S5, applying the new data training model to the business.
2. The retail scene voice interaction optimization method based on knowledge graph according to claim 1, characterized in that in S1, learning is performed by reading corpora, a word stock of itself is established, and dialect recognition problem is solved;
according to the agreement of users, the special calling methods and different accents of commodities in different regions in a wide area are collected anonymously, in a voice shopping guide scene, default voices are related to services, and corpora are collected through machine learning and supervised learning to build a word stock in a vertical industry.
3. The method for retail scene voice interaction optimization based on knowledge graph according to claim 1, wherein in the step S2, a brand knowledge graph is established by listing brand professional words and learning different regional pronunciations and synonyms across the country, and the voice recognition effect can be greatly improved by combining with voice recognition and NLP.
4. The method of claim 3, wherein the accuracies of brand specific word recognition, specific word meaning analysis and scene relationship analysis in S2 are matched with each other.
5. The method of claim 1, wherein the data in the voice-interactive shopping search in S3 is updated weekly, stored in a server, and received and transmitted through a system;
the system performs voice recognition and input through network and personal input.
6. The method of claim 1, wherein the homophones and synonyms in the S4 are listed according to original words, and are divided into homophones, harmonious multiwords and harmonious homowords, and include numbers, arabic letters and a few voices.
7. The method of claim 1, wherein the step of S5 is applied to business, the voice data is transmitted to a voice system, and the data is stored for application, and the steps of S1 are looped.
8. The method of claim 1, wherein the system for recognizing the complete isolated word from the specific person in S1 generally comprises inputting speech, preprocessing speech signals, extracting features, training and recognizing;
the process of speech recognition can be regarded as the process of pattern matching, the process of pattern matching refers to the process of making unknown pattern and a certain model in the model base obtain the best matching according to certain criteria, the reference template that needs to be used in the pattern matching is obtained through template training, in the training stage, after certain processing is carried out on the characteristic parameters, a model is established for each entry and is stored as the template base, in the recognition stage, speech signals pass through the same channel to obtain speech characteristic parameters, a test template is generated and is matched with the reference template, the reference template with the highest matching score is used as the recognition result, and meanwhile, the accuracy of recognition can be improved with the help of a priori knowledge.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011179681.4A CN112259102A (en) | 2020-10-29 | 2020-10-29 | Retail scene voice interaction optimization method based on knowledge graph |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011179681.4A CN112259102A (en) | 2020-10-29 | 2020-10-29 | Retail scene voice interaction optimization method based on knowledge graph |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112259102A true CN112259102A (en) | 2021-01-22 |
Family
ID=74262823
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011179681.4A Pending CN112259102A (en) | 2020-10-29 | 2020-10-29 | Retail scene voice interaction optimization method based on knowledge graph |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112259102A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113076427A (en) * | 2021-03-16 | 2021-07-06 | 海信视像科技股份有限公司 | Media resource searching method, display equipment and server |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106909662A (en) * | 2017-02-27 | 2017-06-30 | 腾讯科技（上海）有限公司 | Knowledge mapping construction method and device |
| CN106971721A (en) * | 2017-03-29 | 2017-07-21 | 沃航（武汉）科技有限公司 | A kind of accent speech recognition system based on embedded mobile device |
| CN109002516A (en) * | 2018-07-06 | 2018-12-14 | 国网电子商务有限公司 | A kind of searching method and device |
| CN111046161A (en) * | 2019-12-19 | 2020-04-21 | 苏州思必驰信息科技有限公司 | Intelligent dialogue method and device for commodity marketing scene |
| CN111354363A (en) * | 2020-02-21 | 2020-06-30 | 镁佳(北京)科技有限公司 | Vehicle-mounted voice recognition method and device, readable storage medium and electronic equipment |
| CN111837116A (en) * | 2017-12-18 | 2020-10-27 | 财富智慧股份有限公司 | Method, computer arrangement and computer-readable storage medium for automatically building or updating a hierarchical dialog flow management model for a conversational AI agent system |
-
2020
- 2020-10-29 CN CN202011179681.4A patent/CN112259102A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106909662A (en) * | 2017-02-27 | 2017-06-30 | 腾讯科技（上海）有限公司 | Knowledge mapping construction method and device |
| CN106971721A (en) * | 2017-03-29 | 2017-07-21 | 沃航（武汉）科技有限公司 | A kind of accent speech recognition system based on embedded mobile device |
| CN111837116A (en) * | 2017-12-18 | 2020-10-27 | 财富智慧股份有限公司 | Method, computer arrangement and computer-readable storage medium for automatically building or updating a hierarchical dialog flow management model for a conversational AI agent system |
| CN109002516A (en) * | 2018-07-06 | 2018-12-14 | 国网电子商务有限公司 | A kind of searching method and device |
| CN111046161A (en) * | 2019-12-19 | 2020-04-21 | 苏州思必驰信息科技有限公司 | Intelligent dialogue method and device for commodity marketing scene |
| CN111354363A (en) * | 2020-02-21 | 2020-06-30 | 镁佳(北京)科技有限公司 | Vehicle-mounted voice recognition method and device, readable storage medium and electronic equipment |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113076427A (en) * | 2021-03-16 | 2021-07-06 | 海信视像科技股份有限公司 | Media resource searching method, display equipment and server |
| CN113076427B (en) * | 2021-03-16 | 2023-02-28 | 海信视像科技股份有限公司 | Media resource searching method, display equipment and server |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10657969B2 (en) | Identity verification method and apparatus based on voiceprint | |
| CN107993665B (en) | Method for determining role of speaker in multi-person conversation scene, intelligent conference method and system | |
| RU2672176C2 (en) | Natural expression processing method, processing and response method, device and system | |
| CN107657017A (en) | Method and apparatus for providing voice service | |
| CN110459210A (en) | Answering method, device, equipment and storage medium based on speech analysis | |
| CN109658271A (en) | A kind of intelligent customer service system and method based on the professional scene of insurance | |
| CN109949799B (en) | Semantic parsing method and system | |
| CN109582788A (en) | Comment spam training, recognition methods, device, equipment and readable storage medium storing program for executing | |
| CN108735200A (en) | A kind of speaker's automatic marking method | |
| Kopparapu | Non-linguistic analysis of call center conversations | |
| CN112233680A (en) | Speaker role identification method and device, electronic equipment and storage medium | |
| CN113254613A (en) | Dialogue question-answering method, device, equipment and storage medium | |
| CN116092472A (en) | Speech synthesis method and synthesis system | |
| CN113051380A (en) | Information generation method and device, electronic equipment and storage medium | |
| CN110852075B (en) | Voice transcription method and device capable of automatically adding punctuation marks and readable storage medium | |
| CN113836945B (en) | Intention recognition method, device, electronic equipment and storage medium | |
| CN112259102A (en) | Retail scene voice interaction optimization method based on knowledge graph | |
| CN112231440A (en) | Voice search method based on artificial intelligence | |
| CN116129868A (en) | Method and system for generating structured photo | |
| CN113505606B (en) | Training information acquisition method and device, electronic equipment and storage medium | |
| CN113850290B (en) | Text processing and model training method, device, equipment and storage medium | |
| CN115831125A (en) | Speech recognition method, device, equipment, storage medium and product | |
| CN115022471A (en) | Intelligent robot voice interaction system and method | |
| CN114519094A (en) | Method and device for conversational recommendation based on random state and electronic equipment | |
| CN114220425A (en) | Chat robot system and conversation method based on voice recognition and Rasa framework |
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 |