CN106649258A - Intelligent question and answer system - Google Patents
Intelligent question and answer system Download PDFInfo
- Publication number
- CN106649258A CN106649258A CN201610844734.7A CN201610844734A CN106649258A CN 106649258 A CN106649258 A CN 106649258A CN 201610844734 A CN201610844734 A CN 201610844734A CN 106649258 A CN106649258 A CN 106649258A
- Authority
- CN
- China
- Prior art keywords
- answer
- key
- module
- ranking
- answers
- 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
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/253—Grammatical analysis; Style critique
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
- G06N5/022—Knowledge engineering; Knowledge acquisition
Abstract
The invention discloses an intelligent question and answer system. In the system, a content obtaining module is used for confirming and collecting answer-related contents, analyzing and classifying questions of question spaces, and automatically expanding a database; a question analysis module is used for analyzing input question information, determining question types, discovering relationships between the questions, and decomposing the questions; a hypothesis generation module is used for searching for the answer-related contents as reference answers as more as possible in a data source; a soft filtering module performs screening on a large amount of the reference answers to obtain a category of the reference answers which are correct answers very possibly, and a category of the reference answers which are the correct answers possibly; an evidence scoring module performs detailed scoring on the reference answers according to a scoring object and determines a degree of closeness to the reference answers; and an answer combining and ranking module combines split answers, calculates credibility and performs ranking, and the highest-ranked answers are optimal answers regarded by the system.
Description
Technical field
The application is related to human-computer intellectualization technical field, more particularly to a kind of intelligent question answering system.
Background technology
Automatically request-answering system is referred to natural language understanding technology as core so that computer is it will be appreciated that the talk of user
Content, realizes the effective communication between people and computer, and provides powerful search capability, accurately answers asking for user
Topic.Wherein, the intelligent Answer System for generally adopting in computer customer service system at present is exactly a kind of automatically request-answering system, and it is
One kind passes through natural language technology, it is to be understood that the problem of user, and provides the artificial intelligence system of accurate answer.
The present invention copes with the challenge in Jeopardy, and the match of program is carried out with a kind of unique question and answer mode, asked
What topic was arranged covers face widely, is related to history, literature, art, pop culture, science and technology, physical culture, geography, play with words
Etc. every field.According to the various clues provided in answer form, entrant must be made brief correct in the form of problem
Return.With general information please conversely, puing question in answer form in Jeopardy, enquirement form is answered.Entrant need to possess and go through
The knowledge such as history, literature, politics, science and pop culture, must parse obscure implication, irony and riddle etc., and computer is not
Being good at carries out this kind of complicated thinking.This intelligent Answer System can be good at meeting the logical thinking, and possess from magnanimity
The ability of correct option is found in data source, also marking ranking is carried out to the confidence level of Key for Reference.
The content of the invention
For achieving the above object, the technical solution used in the present invention is a kind of intelligent question answering system, and the system includes interior
Hold acquisition module, case study module, assume generation module, soft filtering module, evidence scoring modules, answer merging and ranking mould
Block.
Step one, content obtaining module, for confirming and collecting the content related to answer, enter to the problem of problem space
Row is analyzed and classified, and data bank is expanded automatically；
Step 2, case study module, for the problem information of analysis input, determine problem types, realize between problem
Relation and resolution problem；
Step 3, hypothesis generation module, for the content conduct related to answer of the search as much as possible from data source
Key for Reference；
Step 4, soft filtering module, screen to substantial amounts of Key for Reference, separate the class for being likely to correct option
With the class for being likely to be correct option；
Step 5, evidence scoring modules, detailed marking is carried out according to marking object to Key for Reference, determines its close ginseng
Examine the degree of answer；
Step 6, answer merging and ranking module, the answer of partition is merged, and calculates confidence level, and ranking, and ranking is most
Forward is the optimum answer that system thinks.
Description of the drawings
Fig. 1 intelligent Answer System framework schematic diagrams.
Fig. 2 intelligent Answer System flow processs Fig. 1.
Fig. 3 intelligent Answer System flow processs Fig. 2.
Fig. 4 answers merging and ranking model.
Specific embodiment
Below in conjunction with the accompanying drawings the present invention is further elaborated with specific embodiment.
Overall system architecture is made up of four parts：Case study, search, Answer extracting and database, as shown in Figure 1.
For the further decomposition of system, as shown in Figure 2.For the problem of the system that is input to, particular flow sheet is as shown in figure 3, specifically such as
Under：
S301, the system have a graphical interfaces, and designing user input frame carrys out the problem of receiving user's input.
S302, problem is analyzed, the type of decision problem, the later stage can take different retrievals according to the type of problem
And scoring scheme, and problem is automatically added in corresponding database, database is expanded automatically.
S303, several minor issues are segmented into for some long problems that can divide, are extracted according to grammer and word segmentation result
Go out multiple problems.
S304, associated answer is searched for from database for each minor issue for separating, it is related as Key for Reference
The calculating of property is as shown in formula 1, formula 2.
S305, Key for Reference is given a mark, the high marking of correlation is higher, the method such as formula 3, formula of marking
Shown in 4.
S306, fraction is judged, if greater than threshold value, just jump directly to the ranking stage, if less than threshold value, also
Need to carry out S307 steps.Reason is that the scoring method of S305 is little with the relation of context, it is possible that the reliability to answer
Property is not high.
S306, the position of the data source being located according to Key for Reference carry out context detection, and matching is compared, comprehensive other
The popularity of factor, such as data, reliability degree.Give a mark again.
S306, weights are given according to certain weight computing to each fraction, summation draws the total score of Key for Reference.
S307, can all there is a corresponding fraction to all of Key for Reference here, according to the height ranking of fraction
S308, confidence level is calculated according to algorithm, with reference to S307, detailed process is as shown in figure 4, machine learning can be used
Method trains model, and confidence level calculating is carried out automatically.
Content obtaining module is specifically included：For towards field classification classification is carried out to the type of problem, summary is appeared
To field characteristic, the content related to answer is searched for from various texts.By problem participle, t is designated asi, participle tiAt one
The fraction of data source is designated as pi, when in text include participle ti, wij=idf (tj)；Otherwise wij=0.
Wherein,
C (t) represents the number of files comprising participle t, and N represents the number of the All Files in data source.
The case study module is specifically included：Confirm problem types, problem is classified, ask for different types of
Topic has different processing methods, and realizes the relation between problem, then resolution problem.
The relationship module realized between problem is specifically included：The problem of input is compared with the problem in database
Compared with, phraseological SVO relation and semantically contacting between Mining Problems, some problem answers are directly produced from this step.
The resolution problem module is specifically included：Answer is faster more accurately found by resolution problem, by a complexity
The problem of clause is divided into multiple simple questions, parallel processing each problem, and the Feasible degree marking of respectively answer.
The hypothesis generation module is specifically included：The search as much as possible content related to answer, pin from data source
Different searching algorithms is used different types of problem, and all related contents are all as Key for Reference.
The soft filtering module is specifically included：Key for Reference is screened with the marking algorithm of lightweight, by mistake
The Key for Reference of filter needs to enter scoring modules, is not entered by the Key for Reference of filter and merges ranking module.
The scoring modules are specifically included：Find support reference from the context into the Key for Reference of scoring modules to answer
The Additional evidence of case, gives a mark according to the semantic, contact of grammer and various marking algorithms to Key for Reference, more possibly refers to
Answer marking is higher.The object of marking is position, word support, popularity, information reliability.Concrete grammar is：
Two one-dimension arrays P, Q, a two-dimensional array score are created first；The symbol of participle, score are deposited in P and Q
Middle storage fraction simultaneously initializes score [i] [j]=0；
Then each score [i] [j] is calculated, with below equation
Wherein
if t1=t2
Especially sim (FOCUS, CANDIDATE)=log (N).
The merging ranking module is specifically included：By the problem fraction for splitting according to certain weight number combining, calculate total
Point；Confidence is trained by machine learning algorithm and estimates model, problem-targeted total score automatically generates Confidence estimation, side by side
Name.
Problem to being input into is processed, and Key for Reference is scanned in a variety of data sources, and to reference
Answer is split, is given a mark, merging treatment, finally obtains the ranking and confidence level of Key for Reference, exports answer.
Claims (10)
1. a kind of intelligent question answering system, it is characterised in that：The system includes content obtaining module, case study module, assumes
Generation module, soft filtering module, evidence scoring modules, answer merging and ranking module；
Step one, content obtaining module, for confirming and collecting the content related to answer, are carried out point to the problem of problem space
Analyse and classify, data bank is expanded automatically；
Step 2, case study module, for the problem information of analysis input, determine problem types, the relation realized between problem
And resolution problem；
Step 3, hypothesis generation module, for the content related to answer of the search as much as possible from data source as reference
Answer；
Step 4, soft filtering module, screen to substantial amounts of Key for Reference, separate and are likely to a class of correct option and have
A possibly class of correct option；
Step 5, evidence scoring modules, detailed marking is carried out according to marking object to Key for Reference, determines that its close reference is answered
The degree of case；
Step 6, answer merging and ranking module, the answer of partition is merged, and calculates confidence level, and ranking, and ranking is most forward
Be the optimum answer thought of system.
2. a kind of intelligent question answering system according to claim 1, it is characterised in that：For the problem of the system that is input to,
It is specific as follows：
S301, the system have a graphical interfaces, and designing user input frame carrys out the problem of receiving user's input；
S302, problem is analyzed, the type of decision problem, the later stage can take different retrieval and beat according to the type of problem
Offshoot program, and problem is automatically added in corresponding database, database is expanded automatically；
S303, for some long problems that can divide are segmented into several minor issues, extracted according to grammer and word segmentation result many
Individual problem；
S304, associated answer is searched for from database for each minor issue for separating, as Key for Reference；
S305, Key for Reference is given a mark, the high marking of correlation is higher；
S306, fraction is judged, if greater than threshold value, the ranking stage is just jumped directly to, if less than threshold value, in addition it is also necessary to
Carry out S307 steps；Reason be the relation of scoring method and the context of S305 less, it is possible that to the reliability of answer not
It is high；
S306, the position of data source being located according to Key for Reference carry out context detection, and matching is compared, it is comprehensive it is other because
The popularity of element, such as data, reliability degree；Give a mark again；
S306, weights are given according to certain weight computing to each fraction, summation draws the total score of Key for Reference；
S307, can all there is a corresponding fraction to all of Key for Reference here, according to the height ranking of fraction
S308, confidence level is calculated according to algorithm, with reference to S307, the method that can use machine learning trains model, enters automatically
Row confidence level is calculated.
3. a kind of intelligent question answering system according to claim 2, it is characterised in that：Content obtaining module is specifically included：
For towards field classification classification is carried out to the type of problem, sum up towards field characteristic, search from various texts
The rope content related to answer；By problem participle, t is designated asi, participle tiThe fraction of data source is designated as p at onei, when bag in text
T containing participlei, wij=idf (tj)；Otherwise wij=0；
Wherein,
C (t) represents the number of files comprising participle t, and N represents the number of the All Files in data source.
4. a kind of intelligent question answering system according to claim 2, it is characterised in that：The case study module is specifically wrapped
Include：Confirm problem types, problem is classified, there are different processing methods for different types of problem, and realize and ask
Relation between topic, then resolution problem.
5. a kind of intelligent question answering system according to claim 2, it is characterised in that：The relation mould realized between problem
Block is specifically included：The problem of input is compared with the problem in database, phraseological SVO relation between Mining Problems
And semantically contact, some problem answers are directly produced from this step.
6. a kind of intelligent question answering system according to claim 2, it is characterised in that：The resolution problem module is specifically wrapped
Include：Answer is faster more accurately found by resolution problem, the problem of a complicated clause is divided into multiple simple questions, and
Row processes each problem, and the Feasible degree marking of respectively answer.
7. a kind of intelligent question answering system according to claim 2, it is characterised in that：The hypothesis generation module is specifically wrapped
Include：The search as much as possible content related to answer, for different types of problem different search is used from data source
Algorithm, all related contents are all as Key for Reference.
8. a kind of intelligent question answering system according to claim 2, it is characterised in that：The soft filtering module is specifically wrapped
Include：Key for Reference is screened with the marking algorithm of lightweight, needs to enter marking mould by the Key for Reference of filter
Block, is not entered by the Key for Reference of filter and merges ranking module.
9. a kind of intelligent question answering system according to claim 2, it is characterised in that：The scoring modules are specifically included：
The Additional evidence for supporting Key for Reference is found from the context into the Key for Reference of scoring modules, according to semantic, grammer
Contact and various marking algorithms are given a mark to Key for Reference, and more possibly Key for Reference marking is higher；The object of marking is position
Put, word support, popularity, information reliability；Concrete grammar is：
Two one-dimension arrays P, Q, a two-dimensional array score are created first；The symbol of participle is deposited in P and Q, is deposited in score
Put fraction and initialize score [i] [j]=0；Then each score [i] [j] is calculated, with below equation
Wherein
if t1=t2
Especially sim (FOCUS, CANDIDATE)=log (N).
10. a kind of intelligent question answering system according to claim 2, it is characterised in that：The merging ranking module is concrete
Including：By the problem fraction for splitting according to certain weight number combining, total score is calculated；Confidence is trained by machine learning algorithm
Degree estimates model, and problem-targeted total score automatically generates Confidence estimation, and ranking；
Problem to being input into is processed, and Key for Reference is scanned in a variety of data sources, and to Key for Reference
Split, given a mark, merging treatment, finally obtain the ranking and confidence level of Key for Reference, export answer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610844734.7A CN106649258A (en) | 2016-09-22 | 2016-09-22 | Intelligent question and answer system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610844734.7A CN106649258A (en) | 2016-09-22 | 2016-09-22 | Intelligent question and answer system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106649258A true CN106649258A (en) | 2017-05-10 |
Family
ID=58853197
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610844734.7A Pending CN106649258A (en) | 2016-09-22 | 2016-09-22 | Intelligent question and answer system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106649258A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107748795A (en) * | 2017-11-03 | 2018-03-02 | 深圳市中润四方信息技术有限公司 | A kind of method, system and device for building knowledge base |
| CN108108449A (en) * | 2017-12-27 | 2018-06-01 | 哈尔滨福满科技有限责任公司 | A kind of implementation method based on multi-source heterogeneous data question answering system and the system towards medical field |
| CN108595494A (en) * | 2018-03-15 | 2018-09-28 | 腾讯科技（深圳）有限公司 | The acquisition methods and device of reply message |
| CN108717413A (en) * | 2018-03-26 | 2018-10-30 | 浙江大学 | It is a kind of based on the assumption that property semi-supervised learning Opening field answering method |
| CN108920554A (en) * | 2018-06-20 | 2018-11-30 | 大国创新智能科技(东莞)有限公司 | Innovative approach and innovative education robot system based on big data and artificial intelligence |
| CN109002540A (en) * | 2018-07-23 | 2018-12-14 | 电子科技大学 | It is a kind of Chinese notice document problem answers to automatic generation method |
| CN109783704A (en) * | 2019-01-03 | 2019-05-21 | 中国科学院自动化研究所 | Man-machine mixed answer method, system, device |
| CN110275949A (en) * | 2019-06-06 | 2019-09-24 | 深圳中兴飞贷金融科技有限公司 | Auto-answer method and system for application of providing a loan |
| CN110309282A (en) * | 2019-06-14 | 2019-10-08 | 北京奇艺世纪科技有限公司 | A kind of answer determines method and device |
| CN110799970A (en) * | 2017-06-27 | 2020-02-14 | 华为技术有限公司 | Question-answering system and question-answering method |
| CN110807087A (en) * | 2019-10-21 | 2020-02-18 | 拉扎斯网络科技（上海）有限公司 | Data processing method and device, readable storage medium and electronic equipment |
| CN111144973A (en) * | 2019-11-29 | 2020-05-12 | 深圳市嘀哒知经科技有限责任公司 | Question ranking method and computer-readable storage medium |
| CN111767374A (en) * | 2019-03-29 | 2020-10-13 | 北京搜狗科技发展有限公司 | Data processing method, device and machine readable medium |
| CN113449117A (en) * | 2021-06-24 | 2021-09-28 | 武汉工程大学 | Bi-LSTM and Chinese knowledge graph-based composite question-answering method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101377777A (en) * | 2007-09-03 | 2009-03-04 | 北京百问百答网络技术有限公司 | Automatic inquiring and answering method and system |
| CN102637192A (en) * | 2012-02-17 | 2012-08-15 | 清华大学 | Method for answering with natural language |
| CN103229168A (en) * | 2010-09-28 | 2013-07-31 | 国际商业机器公司 | Evidence diffusion among candidate answers during question answering |
-
2016
- 2016-09-22 CN CN201610844734.7A patent/CN106649258A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101377777A (en) * | 2007-09-03 | 2009-03-04 | 北京百问百答网络技术有限公司 | Automatic inquiring and answering method and system |
| CN103229168A (en) * | 2010-09-28 | 2013-07-31 | 国际商业机器公司 | Evidence diffusion among candidate answers during question answering |
| CN102637192A (en) * | 2012-02-17 | 2012-08-15 | 清华大学 | Method for answering with natural language |
Non-Patent Citations (1)
| Title |
|---|
| 周澄 等: "Smith-Waterman算法的若干优化及并行实现", 《计算机工程与应用》 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110799970A (en) * | 2017-06-27 | 2020-02-14 | 华为技术有限公司 | Question-answering system and question-answering method |
| CN107748795A (en) * | 2017-11-03 | 2018-03-02 | 深圳市中润四方信息技术有限公司 | A kind of method, system and device for building knowledge base |
| CN108108449A (en) * | 2017-12-27 | 2018-06-01 | 哈尔滨福满科技有限责任公司 | A kind of implementation method based on multi-source heterogeneous data question answering system and the system towards medical field |
| CN108595494A (en) * | 2018-03-15 | 2018-09-28 | 腾讯科技（深圳）有限公司 | The acquisition methods and device of reply message |
| CN108717413A (en) * | 2018-03-26 | 2018-10-30 | 浙江大学 | It is a kind of based on the assumption that property semi-supervised learning Opening field answering method |
| CN108717413B (en) * | 2018-03-26 | 2021-10-08 | 浙江大学 | Open field question-answering method based on hypothetical semi-supervised learning |
| CN108920554A (en) * | 2018-06-20 | 2018-11-30 | 大国创新智能科技(东莞)有限公司 | Innovative approach and innovative education robot system based on big data and artificial intelligence |
| CN108920554B (en) * | 2018-06-20 | 2020-12-22 | 大国创新智能科技(东莞)有限公司 | Creative method based on big data and artificial intelligence and creative education robot system |
| CN109002540B (en) * | 2018-07-23 | 2021-03-16 | 电子科技大学 | Method for automatically generating Chinese announcement document question answer pairs |
| CN109002540A (en) * | 2018-07-23 | 2018-12-14 | 电子科技大学 | It is a kind of Chinese notice document problem answers to automatic generation method |
| CN109783704B (en) * | 2019-01-03 | 2021-02-02 | 中国科学院自动化研究所 | Man-machine hybrid response method, system and device |
| CN109783704A (en) * | 2019-01-03 | 2019-05-21 | 中国科学院自动化研究所 | Man-machine mixed answer method, system, device |
| CN111767374A (en) * | 2019-03-29 | 2020-10-13 | 北京搜狗科技发展有限公司 | Data processing method, device and machine readable medium |
| CN110275949A (en) * | 2019-06-06 | 2019-09-24 | 深圳中兴飞贷金融科技有限公司 | Auto-answer method and system for application of providing a loan |
| CN110309282A (en) * | 2019-06-14 | 2019-10-08 | 北京奇艺世纪科技有限公司 | A kind of answer determines method and device |
| CN110309282B (en) * | 2019-06-14 | 2021-08-27 | 北京奇艺世纪科技有限公司 | Answer determination method and device |
| CN110807087A (en) * | 2019-10-21 | 2020-02-18 | 拉扎斯网络科技（上海）有限公司 | Data processing method and device, readable storage medium and electronic equipment |
| CN111144973A (en) * | 2019-11-29 | 2020-05-12 | 深圳市嘀哒知经科技有限责任公司 | Question ranking method and computer-readable storage medium |
| CN111144973B (en) * | 2019-11-29 | 2023-07-25 | 深圳市嘀哒知经科技有限责任公司 | Question ranking method and computer-readable storage medium |
| CN113449117A (en) * | 2021-06-24 | 2021-09-28 | 武汉工程大学 | Bi-LSTM and Chinese knowledge graph-based composite question-answering method |
| CN113449117B (en) * | 2021-06-24 | 2023-09-26 | 武汉工程大学 | Bi-LSTM and Chinese knowledge graph based compound question-answering method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106649258A (en) | Intelligent question and answer system | |
| CN102804208B (en) | Individual model for visual search application automatic mining famous person | |
| US10565233B2 (en) | Suffix tree similarity measure for document clustering | |
| CN109189901B (en) | Method for automatically discovering new classification and corresponding corpus in intelligent customer service system | |
| CN108764480B (en) | Information processing system | |
| KR101605430B1 (en) | SYSTEM AND METHOD FOR BUINDING QAs DATABASE AND SEARCH SYSTEM AND METHOD USING THE SAME | |
| CN106021410A (en) | Source code annotation quality evaluation method based on machine learning | |
| CN107679110A (en) | The method and device of knowledge mapping is improved with reference to text classification and picture attribute extraction | |
| CN110910283A (en) | Method, device, equipment and storage medium for generating legal document | |
| CN102456060A (en) | Information processing device and information processing method | |
| CN110516259B (en) | Method and device for identifying technical keywords, computer equipment and storage medium | |
| Labusch et al. | Named Entity Disambiguation and Linking Historic Newspaper OCR with BERT. | |
| US20190130030A1 (en) | Generation method, generation device, and recording medium | |
| Kruszewski et al. | Jointly optimizing word representations for lexical and sentential tasks with the c-phrase model | |
| CN110909230A (en) | Network hotspot analysis method and system | |
| Beleveslis et al. | A hybrid method for sentiment analysis of election related tweets | |
| CN116010564A (en) | Construction method of rice pest question-answering system based on multi-mode knowledge graph | |
| Di Giovanni et al. | Content-based stance classification of tweets about the 2020 Italian constitutional referendum | |
| CN116723005A (en) | Method and system for tracking malicious code implicit information under polymorphic hiding | |
| CN111259223A (en) | News recommendation and text classification method based on emotion analysis model | |
| Walha et al. | A Lexicon approach to multidimensional analysis of tweets opinion | |
| CN107784024A (en) | Build the method and device of party's portrait | |
| Otani et al. | Large-scale acquisition of commonsense knowledge via a quiz game on a dialogue system | |
| CN105975480A (en) | Instruction identification method and system | |
| CN115048504A (en) | Information pushing method and device, computer equipment and computer readable storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170510 |
|
| RJ01 | Rejection of invention patent application after publication |