CN107679231A - A kind of vertical field and the implementation method of Opening field mixed type intelligent Answer System - Google Patents
A kind of vertical field and the implementation method of Opening field mixed type intelligent Answer System Download PDFInfo
- Publication number
- CN107679231A CN107679231A CN201711000697.2A CN201711000697A CN107679231A CN 107679231 A CN107679231 A CN 107679231A CN 201711000697 A CN201711000697 A CN 201711000697A CN 107679231 A CN107679231 A CN 107679231A
- Authority
- CN
- China
- Prior art keywords
- field
- corpus
- answering system
- question answering
- sentence
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR 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
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR 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 or dialogue systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Human Computer Interaction (AREA)
- Computational Linguistics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention discloses a kind of vertical field and the implementation method of Opening field mixed type intelligent Answer System, is related to artificial intelligence field；The invention discloses a kind of vertical field and the implementation method of Opening field mixed type intelligent Answer System, language material is deposited in different corpus according to field, corpus is handled, and carry out text vector using word2vec models, vertical domain classification is carried out to the field of problem, specify the field of problem, if so as to be trained for the closing field to question and answer language material using distance algorithm and the problem does not have specific domain features, then question answering system switchs to Opening field, a large amount of corpus, which are trained, using seq2seq deep learnings model makes question answering system automatically generate response.Therefore the automatic switchover in vertical field and Opening field intelligent Answer System can be realized using the inventive method, improves response time and the response degree of accuracy of question answering system.
Description
Technical field
The present invention discloses a kind of implementation method of intelligent Answer System, is related to artificial intelligence field, specifically a kind of
Vertical field and the implementation method of Opening field mixed type intelligent Answer System.
Background technology
Intelligent answer is an important directions of natural language processing field, it is intended to allows user directly to be putd question to natural language
And obtain answer.For example, user's query " where is Baidu mansion ", question answering system answers the " street 10 of ShangDi, Haidian District, BeiJing City ten
Number ".From the point of view of user, intelligent answer is a kind of simple and succinct information acquisition method.User directly uses natural language
Which type of interacted with question answering system, without regard to keyword combination to represent the intention of oneself using, so simply；Question and answer system
System directly returns to the answer of problem, and user from tediously long relevant documentation without oneself finding answer content, so succinctly.
The invention discloses a kind of vertical field and the implementation method of Opening field mixed type intelligent Answer System, by language material
Deposited according to field in different corpus, corpus is handled, and text vector is carried out using word2vec models
Change, vertical domain classification is carried out to the field of problem, specifies the field of problem, so as to make for the closing field to question and answer language material
If being trained with distance algorithm and the problem not having specific domain features, question answering system switchs to Opening field, uses
Seq2seq deep learnings model is trained to a large amount of corpus makes question answering system automatically generate response.Therefore the present invention is utilized
Method can realize the automatic switchover in vertical field and Opening field intelligent Answer System, improve the response time of question answering system
With the response degree of accuracy.
The content of the invention
The present invention is directed to problem of the prior art, there is provided a kind of vertical field and Opening field mixed type intelligent Answer System
Implementation method, have broad application prospects.
A kind of vertical field and the implementation method of Opening field mixed type intelligent Answer System：
Language material is deposited in different corpus according to field, corpus is handled, word2vec models is reused and enters
Compose a piece of writing this vectorization, domain classification is carried out to corpus and establishes model；The problem of being received to question answering system is built using corpus
Vertical model carries out domain classification, specifies the field of problem, uses distance to calculate for the corpus belonging to problem in vertical field
Method, which is trained, establishes model, question answering system is automatically generated response and answers sentence；If the no specific vertical domain features of the problem,
Then problem is switched to Opening field by question answering system, and the corpus of question answering system is trained using seq2seq models and establishes mould
Type, question answering system is automatically generated response and answer sentence.
Described method is trained structure model using NB Algorithm to the corpus of classification, and described in use
Model the problem of being received to question answering system carry out domain classification, judge which field is described problem belong to.
In described method, question answering system receive the problem of affiliated vertical field, then by described problem in the vertical field
The corpus that is added into answering sentence in the vertical field of conventional question sentence in, and one is answered sentence and matches a variety of question sentences as far as possible.
In described method, question answering system receive the problem of affiliated vertical field, use Levenshtein distance algorithms pair
Conventional question sentence in the corpus in the vertical field and answer sentence and be trained and establish model, question answering system is automatically generated response and answer
Sentence.
Language material is deposited according to field with Opening field mixed type intelligent Answer System, the question answering system in a kind of vertical field
It is put in different corpus, corpus is handled, reuses word2vec models and carry out text vector, to corpus
Carry out domain classification and establish model；
The model that the problem of being received to question answering system is established using corpus carries out domain classification, specifies the field of problem, for
Corpus belonging to problem in vertical field is trained using distance algorithm and establishes model, question answering system is automatically generated sound
Response sentence；If problem is switched to Opening field by the problem without specific vertical domain features, question answering system, use
Seq2seq models are trained to the corpus of question answering system and establish model, question answering system is automatically generated response and answer sentence.
Described question answering system is trained structure model using NB Algorithm to the corpus of classification, and uses
The problem of described model receives to question answering system carries out domain classification, judges which field is described problem belong to.
The affiliated vertical field of the problem of described question answering system receives, then ask the conventional of described problem in the vertical field
In the corpus that sentence is added into the vertical field with answering sentence, and one is answered sentence and matches a variety of question sentences as far as possible.
The affiliated vertical field of the problem of question answering system receives, using Levenshtein distance algorithms to the vertical neck
Conventional question sentence in the corpus in domain and answer sentence and be trained and establish model, question answering system is automatically generated response and answer sentence.
Usefulness of the present invention is：
The invention discloses a kind of vertical field and the implementation method of Opening field mixed type intelligent Answer System, by language material according to
Field is deposited in different corpus, and corpus is handled, and carries out text vector using word2vec models, right
The field of problem carries out vertical domain classification, specifies the field of problem, so as to for the closing field to the use of question and answer language material away from
If being trained from algorithm and the problem not having specific domain features, question answering system switchs to Opening field, uses
Seq2seq deep learnings model is trained to a large amount of corpus makes question answering system automatically generate response.Therefore the present invention is utilized
Method can realize the automatic switchover in vertical field and Opening field intelligent Answer System, improve the response time of question answering system
With the response degree of accuracy.
Brief description of the drawings
Fig. 1 is the inventive method schematic flow sheet.
Embodiment
The present invention provides a kind of vertical field and the implementation method of Opening field mixed type intelligent Answer System：
Language material is deposited in different corpus according to field, corpus is handled, word2vec models is reused and enters
Compose a piece of writing this vectorization, domain classification is carried out to corpus and establishes model；The problem of being received to question answering system is built using corpus
Vertical model carries out domain classification, specifies the field of problem, uses distance to calculate for the corpus belonging to problem in vertical field
Method, which is trained, establishes model, question answering system is automatically generated response and answers sentence；If the no specific vertical domain features of the problem,
Then problem is switched to Opening field by question answering system, and the corpus of question answering system is trained using seq2seq models and establishes mould
Type, question answering system is automatically generated response and answer sentence.
A kind of vertical field corresponding with the above method and Opening field mixed type intelligent Answer System are provided simultaneously.
Using the inventive method, following steps are substantially carried out, obtain vertical field and Opening field mixed type intelligent answer
System.
Wherein language material is handled：Language material is deposited in different corpus according to field, the related language such as physical culture
Material, is deposited in physical culture corpus, and the related language material of video display is deposited in video display corpus etc.,
Corpus is segmented again, removes stop words etc., text vector is carried out using word2vec models, corpus is entered
Row domain classification simultaneously establishes model using NB Algorithm to the corpus of classification；
Problem searching classification：For question answering system receive user the problem of, use naive Bayesian build corpus model
Problem is classified, judges the problem of it belongs to which field；
Vertical field data retrieval：, will be described in the vertical field if affiliated the problem of question answering system receives is vertical field
In the corpus that the conventional question sentence of problem is added into the vertical field with answering sentence, and one is answered sentence and matches a variety of ask as far as possible
Sentence, the problem of for user, using Levenshtein distance algorithms to the conventional question sentence in the corpus in the vertical field with answering
Sentence, which is trained, establishes model, question answering system is automatically generated response and answers sentence；
And Opening field：If the problem of question answering system receives turns problem without specific vertical domain features, question answering system
For Opening field, the problem of for Opening field, the question and answer language material of Opening field is entered using the seq2seq models of deep learning
Row training modeling, and automatically generate answer.
Therefore the present invention can both apply self-service customer service in vertical closing field, such as bank, the self-help shopping in market,
Self-service phone of enterprise etc. can apply the chat formula question and answer that Opening field is carried out in Opening field again, and ensure quick and precisely
Under conditions of response, response time and the response degree of accuracy of question answering system are improved.
Claims (8)
1. a kind of vertical field and the implementation method of Opening field mixed type intelligent Answer System, it is characterized in that
Language material is deposited in different corpus according to field, corpus is handled, word2vec models is reused and enters
Compose a piece of writing this vectorization, domain classification is carried out to corpus and establishes model；The problem of being received to question answering system is built using corpus
Vertical model carries out domain classification, specifies the field of problem, uses distance to calculate for the corpus belonging to problem in vertical field
Method, which is trained, establishes model, question answering system is automatically generated response and answers sentence；If the no specific vertical domain features of the problem,
Then problem is switched to Opening field by question answering system, and the corpus of question answering system is trained using seq2seq models and establishes mould
Type, question answering system is automatically generated response and answer sentence.
2. according to the method for claim 1, it is characterized in that being instructed using NB Algorithm to the corpus of classification
The problem of white silk builds model, and the model described in use receives to question answering system carries out domain classification, judges that described problem belongs to
Which field.
3. method according to claim 1 or 2, it is characterized in that vertical field belonging to described problem, then by the vertical field
In the corpus that the conventional question sentence of interior described problem is added into the vertical field with answering sentence, and one answer sentence as far as possible match it is more
Kind question sentence.
4. according to the method for claim 3, it is characterized in that vertical field belonging to described problem, using Levenshtein away from
To the conventional question sentence in the corpus in the vertical field and answer sentence from algorithm and be trained and establish model, make question answering system from movable property
Sentence is answered in raw response.
5. a kind of vertical field and Opening field mixed type intelligent Answer System, it is characterized in that the question answering system by language material according to
Field is deposited in different corpus, and corpus is handled, and is reused word2vec models and is carried out text vector, right
Corpus carries out domain classification and establishes model；
The model that the problem of being received to question answering system is established using corpus carries out domain classification, specifies the field of problem, for
Corpus belonging to problem in vertical field is trained using distance algorithm and establishes model, question answering system is automatically generated sound
Response sentence；If problem is switched to Opening field by the problem without specific vertical domain features, question answering system, use
Seq2seq models are trained to the corpus of question answering system and establish model, question answering system is automatically generated response and answer sentence.
6. question answering system according to claim 5, it is characterized in that the question answering system uses NB Algorithm to dividing
The problem of corpus of class is trained structure model, and the model described in use receives to question answering system carries out domain classification,
Judge which field is described problem belong to.
7. the question answering system according to claim 5 or 6, vertically led it is characterized in that the question answering system is affiliated the problem of reception
Domain, then in the corpus being added into the conventional question sentence of described problem in the vertical field in the vertical field with answering sentence, and
Answer sentence for one and match a variety of question sentences as far as possible.
8. question answering system according to claim 7, it is characterized in that the affiliated vertical field of the problem of question answering system receives,
To the conventional question sentence in the corpus in the vertical field and answer sentence using Levenshtein distance algorithms and be trained and establish mould
Type, question answering system is automatically generated response and answer sentence.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711000697.2A CN107679231A (en) | 2017-10-24 | 2017-10-24 | A kind of vertical field and the implementation method of Opening field mixed type intelligent Answer System |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711000697.2A CN107679231A (en) | 2017-10-24 | 2017-10-24 | A kind of vertical field and the implementation method of Opening field mixed type intelligent Answer System |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107679231A true CN107679231A (en) | 2018-02-09 |
Family
ID=61141341
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711000697.2A Pending CN107679231A (en) | 2017-10-24 | 2017-10-24 | A kind of vertical field and the implementation method of Opening field mixed type intelligent Answer System |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107679231A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109002515A (en) * | 2018-07-04 | 2018-12-14 | 网宿科技股份有限公司 | A kind of method and apparatus of intelligent response |
| CN109977207A (en) * | 2019-03-21 | 2019-07-05 | 网易（杭州）网络有限公司 | Talk with generation method, dialogue generating means, electronic equipment and storage medium |
| CN110543636A (en) * | 2019-09-06 | 2019-12-06 | 出门问问(武汉)信息科技有限公司 | training data selection method of dialogue system |
| CN110826341A (en) * | 2019-11-26 | 2020-02-21 | 杭州微洱网络科技有限公司 | Semantic similarity calculation method based on seq2seq model |
| CN111339254A (en) * | 2020-02-26 | 2020-06-26 | 常州市贝叶斯智能科技有限公司 | Intelligent voice processing method and device, intelligent equipment and medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101320374A (en) * | 2008-07-10 | 2008-12-10 | 昆明理工大学 | Field question classification method combining syntax structural relationship and field characteristic |
| CN101373532A (en) * | 2008-07-10 | 2009-02-25 | 昆明理工大学 | FAQ Chinese request-answering system implementing method in tourism field |
| CN103902672A (en) * | 2014-03-19 | 2014-07-02 | 微梦创科网络科技（中国）有限公司 | Question answering system and dealing method thereof |
| CN104156469A (en) * | 2014-08-22 | 2014-11-19 | 百度在线网络技术（北京）有限公司 | Data acquisition method and device |
| CN105068661A (en) * | 2015-09-07 | 2015-11-18 | 百度在线网络技术（北京）有限公司 | Man-machine interaction method and system based on artificial intelligence |
| US20150371137A1 (en) * | 2014-06-19 | 2015-12-24 | International Business Machines Corporation | Displaying Quality of Question Being Asked a Question Answering System |
| CN105843897A (en) * | 2016-03-23 | 2016-08-10 | 青岛海尔软件有限公司 | Vertical domain-oriented intelligent question and answer system |
| CN106919646A (en) * | 2017-01-18 | 2017-07-04 | 南京云思创智信息科技有限公司 | Chinese text summarization generation system and method |
| CN106997375A (en) * | 2017-02-28 | 2017-08-01 | 浙江大学 | Recommendation method is replied in customer service based on deep learning |
| CN107169105A (en) * | 2017-05-17 | 2017-09-15 | 北京品智能量科技有限公司 | Question and answer system and method for vehicle |
-
2017
- 2017-10-24 CN CN201711000697.2A patent/CN107679231A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101320374A (en) * | 2008-07-10 | 2008-12-10 | 昆明理工大学 | Field question classification method combining syntax structural relationship and field characteristic |
| CN101373532A (en) * | 2008-07-10 | 2009-02-25 | 昆明理工大学 | FAQ Chinese request-answering system implementing method in tourism field |
| CN103902672A (en) * | 2014-03-19 | 2014-07-02 | 微梦创科网络科技（中国）有限公司 | Question answering system and dealing method thereof |
| US20150371137A1 (en) * | 2014-06-19 | 2015-12-24 | International Business Machines Corporation | Displaying Quality of Question Being Asked a Question Answering System |
| CN104156469A (en) * | 2014-08-22 | 2014-11-19 | 百度在线网络技术（北京）有限公司 | Data acquisition method and device |
| CN105068661A (en) * | 2015-09-07 | 2015-11-18 | 百度在线网络技术（北京）有限公司 | Man-machine interaction method and system based on artificial intelligence |
| CN105843897A (en) * | 2016-03-23 | 2016-08-10 | 青岛海尔软件有限公司 | Vertical domain-oriented intelligent question and answer system |
| CN106919646A (en) * | 2017-01-18 | 2017-07-04 | 南京云思创智信息科技有限公司 | Chinese text summarization generation system and method |
| CN106997375A (en) * | 2017-02-28 | 2017-08-01 | 浙江大学 | Recommendation method is replied in customer service based on deep learning |
| CN107169105A (en) * | 2017-05-17 | 2017-09-15 | 北京品智能量科技有限公司 | Question and answer system and method for vehicle |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109002515A (en) * | 2018-07-04 | 2018-12-14 | 网宿科技股份有限公司 | A kind of method and apparatus of intelligent response |
| CN109977207A (en) * | 2019-03-21 | 2019-07-05 | 网易（杭州）网络有限公司 | Talk with generation method, dialogue generating means, electronic equipment and storage medium |
| CN110543636A (en) * | 2019-09-06 | 2019-12-06 | 出门问问(武汉)信息科技有限公司 | training data selection method of dialogue system |
| CN110826341A (en) * | 2019-11-26 | 2020-02-21 | 杭州微洱网络科技有限公司 | Semantic similarity calculation method based on seq2seq model |
| CN111339254A (en) * | 2020-02-26 | 2020-06-26 | 常州市贝叶斯智能科技有限公司 | Intelligent voice processing method and device, intelligent equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107679231A (en) | A kind of vertical field and the implementation method of Opening field mixed type intelligent Answer System | |
| CN107329967B (en) | Question answering system and method based on deep learning | |
| CN106056207B (en) | A kind of robot depth interaction and inference method and device based on natural language | |
| CN107862000B (en) | A kind of agricultural technology consulting interactive method | |
| CN106649524B (en) | A kind of deep learning intelligent response system of the modified based on computer cloud data | |
| CN107818164A (en) | A kind of intelligent answer method and its system | |
| CN108108849A (en) | A kind of microblog emotional Forecasting Methodology based on Weakly supervised multi-modal deep learning | |
| CN109271537B (en) | Text-to-image generation method and system based on distillation learning | |
| CN107562863A (en) | Chat robots reply automatic generation method and system | |
| CN103761975B (en) | Method and device for oral evaluation | |
| CN107918633B (en) | Sensitive public opinion content identification method and early warning system based on semantic analysis technology | |
| CN111081220B (en) | Vehicle-mounted voice interaction method, full-duplex dialogue system, server and storage medium | |
| CN109036391A (en) | Audio recognition method, apparatus and system | |
| CN106991161A (en) | A kind of method for automatically generating open-ended question answer | |
| CN106503123A (en) | A kind of deep learning intelligent response system based on computer cloud data | |
| CN107544956B (en) | Text key point detection method and system | |
| CN109558576A (en) | A kind of punctuation mark prediction technique based on from attention mechanism | |
| CN110825850B (en) | Natural language theme classification method and device | |
| CN112559749B (en) | Intelligent matching method, device and storage medium for online education teachers and students | |
| CN109360458A (en) | Interest assistant teaching method, device and robot | |
| CN109885661A (en) | Educate the question answering system under scene | |
| CN109325780A (en) | A kind of exchange method of the intelligent customer service system in E-Governance Oriented field | |
| CN109064389A (en) | A kind of free hand line draws the deep learning method of generation presence image | |
| CN110245253A (en) | A kind of Semantic interaction method and system based on environmental information | |
| CN112132536A (en) | Post recommendation method, system, computer equipment and 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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180209 |