CN106126502A - A kind of emotional semantic classification system and method based on support vector machine - Google Patents
A kind of emotional semantic classification system and method based on support vector machine Download PDFInfo
- Publication number
- CN106126502A CN106126502A CN201610529672.0A CN201610529672A CN106126502A CN 106126502 A CN106126502 A CN 106126502A CN 201610529672 A CN201610529672 A CN 201610529672A CN 106126502 A CN106126502 A CN 106126502A
- Authority
- CN
- China
- Prior art keywords
- text
- word
- feature
- support vector
- vector machine
- 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.)
- Granted
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
- G06F40/237—Lexical tools
- G06F40/247—Thesauruses; Synonyms
-
- 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
- G06F16/34—Browsing; Visualisation therefor
-
- 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
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/374—Thesaurus
-
- 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
Abstract
The present invention relates to the analysis of public opinion technology, it discloses a kind of emotional semantic classification system and method based on support vector machine, from user comment information, find public sentiment for quick, accurate.The present invention utilizes reptile module to obtain user and is published in the review information of forum, by data being carried out the pretreatment such as participle, obtain the feature phrase of comment text and there is the training data of typicality, subsequently training data is carried out Emotion tagging, and utilize support vector machine that training data is calculated, obtain disaggregated model, according to disaggregated model, evaluation text to be sorted is analyzed, obtain the affective state estimated, finally utilize visualization model, show classification results, user is helped quickly to understand user feeling based on different entities object (keyword), and and then understand internet public feelings, it is applicable to website, the analysis of public opinion of forum.
Description
Technical field
The present invention relates to the analysis of public opinion technology, be specifically related to a kind of emotional semantic classification system based on support vector machine and side
Method.
Background technology
Along with the fast development of the Internet, the data on the Internet present explosive growth.According to incompletely statistics, 1 minute
In, the upper newly-increased microblogging of Twitter reaches 100,000.And at home, Sina's microblog users number 6.5 hundred million, day any active ues reach 4600
Ten thousand, Tengxun's microblog users number 6.2 hundred million, day any active ues about 100,000,000；Moreover, valuable information in traditional forum website
About 1 year about 100,000,000.The hugest any active ues and abundant in content, the comment that emotion is distinct issued thereof behind,
The most numerous valuable information.Analysis to these information, can help to find commentator's emotion to special body, example
As: microblogging/forum user, for enterprise " front " or the evaluation of " negatively ", for the viewpoint etc. of social group's event, thus is helped
Help others grasp spin, problem analysis cause etc..
But, comment text is classified, and finds that the emotion preference of user is a challenging job, example
As: certain user A has delivered the model of " conwoman that telecommunications worker is pretended to be in attention ", and user B replys and says that " money of old man is good
Deceive." discounting for the scene of text, only sentence itself is carried out emotion differentiation, often obtain inconsistent judged result.
To this end, we have developed a kind of sensibility classification method based on support vector machine, for user is published in microblogging, forum
Text message is classified, and then analyzes the public sentiment situation for special body.
Summary of the invention
The technical problem to be solved is: propose a kind of emotional semantic classification system based on support vector machine and side
Method, finds public sentiment for quick, accurate from user comment information.
The technical solution adopted for the present invention to solve the technical problems is:
A kind of emotional semantic classification system based on support vector machine, comprising:
Data acquisition and pretreatment module, be responsible for utilizing web crawlers to carry out data and crawl, and what acquisition user was delivered comments
Opinion information, and review information is carried out pretreatment；
Feature Words and training sample generation module, be responsible for using the comment text through pretreatment as input, choose with
The high frequency words of specific part of speech is as Feature Words, and adds feature dictionary；Choose the evaluation text comprising Feature Words as training sample
This, and the emotion of training sample is manually marked；
Svm classifier module, is responsible for based on feature dictionary, and training sample extracts characteristic vector, and vector is supported in input
Machine generates disaggregated model；Utilize disaggregated model that the to be sorted emotion value evaluating text is calculated, analyze the emotion of text
Orientation；
Visualization model, is responsible for representing analysis result in web terminal.
Additionally, present invention also offers a kind of sensibility classification method based on support vector machine, it comprises the following steps:
A, utilize web crawlers to carry out data to crawl, obtain the review information that user is delivered, and review information is carried out
Pretreatment；
B, using the comment text through pretreatment as input, choose the high frequency words with specific part of speech as Feature Words,
And add feature dictionary；Choose the evaluation text comprising Feature Words as training sample, and the emotion of training sample is carried out people
Work marks；
C, based on feature dictionary, to training sample extract characteristic vector, input support vector machine generate disaggregated model；
Utilize disaggregated model that the to be sorted emotion value evaluating text is calculated, analyze the orientation of emotion of text；
D, analysis result is represented in web terminal.
As optimizing further, in step A, described utilize web crawlers to carry out data to crawl, obtain what user was delivered
Review information, specifically includes:
From the beginning of the website specified, crawling webpage with the pattern of breadth-first, the webpage got for each, to it
Page source code resolves, and obtains user comment information in webpage, the review information write into Databasce that will obtain.
As optimizing further, in step A, described review information is carried out pretreatment, specifically includes:
Use Chinese word segmentation tool kit that the evaluation information of user is carried out participle, and mark part of speech.
As optimizing further, in step B, described in choose the high frequency words with specific part of speech as Feature Words, specifically wrap
Include: be that noun, verb and adjectival frequent words are as Feature Words based on FindCover algorithm picks part of speech.
As optimizing further, described is noun, verb and adjectival high frequency based on FindCover algorithm picks part of speech
Word as Feature Words, method particularly includes:
Determine the input of FindCover algorithm: participle also marks the evaluation text collection U of part of speech, Feature Words number n, spy
Levy word length L, part of speech set P；
Determine the output of FindCover algorithm: feature phrase S；
The process of choosing includes:
Step 1, initialization set S, A；
Step 2, calculating mapping relations Map M, be mapped to one group of text id:M comprising this word by each word word
(word)；
Step 3, when gathering S and not comprising n word, then find word word so that it is satisfied three conditions:
I () part of speech meets the requirement of P；
(ii) length meets the requirement of L；
(iii) current coverage rate coverage=| M (word)-A | is maximum；
If coverage rate coverage=0 of the word that step 4 searches out, then terminate circulation, otherwise, word is added
S, adds A by M (word), returns step 3 and continues cycling through, until set S comprises n word or the coverage rate of word searched out
Coverage=0；
Step 5, return set S are as feature phrase.
As optimizing further, the value of described n, P, L can be adjusted according to practical situation.
As optimizing further, in step B, described in choose the evaluation text comprising Feature Words as training sample, specifically
Including:
The feature phrase S returned according to FindCover algorithm, uses following strategy to choose training sample: first, exports institute
There is the evaluation text collection U comprising Feature WordsfIf, | Uf| > 1% | U |, then from UfIn randomly choose 1% | U | individual evaluation text make
For training sample；Otherwise export UfAs training sample.
As optimizing further, in step C, described characteristic vector that training sample is extracted, input support vector machine generation
Disaggregated model, specifically includes:
First according to Feature Words, the text in sample data is converted to shape such as "<labelling>feature 1: number feature 2: individual
Number ... feature n: number " form, according to three way classification, then<labelling>value be positive, negative or
neutral；According to two way classification, then<labelling>value is positive and negative；The training data will changed subsequently
It is input in LIBSVM storehouse carry out classification based training.
As optimizing further, in step D, analysis result is represented in web terminal, described in the content that represents include:
" front ", " negatively ", the ratio of " neutral " of text based on particular keywords, the urtext that emotion is relevant, temporally tie up
Degree represents the emotion change of text.
The invention has the beneficial effects as follows: utilize reptile module to obtain user and be published in the review information of forum, pass through logarithm
According to carrying out the pretreatment such as participle, obtain the feature phrase of comment text and there is the training data of typicality, subsequently to training
Data carry out Emotion tagging, and utilize support vector machine to calculate training data, obtain disaggregated model, according to classification mould
Type, is analyzed evaluation text to be sorted, obtains the affective state estimated, finally utilizes visualization model, shows classification
As a result, help user quickly to understand user feeling based on different entities object (keyword), and and then understand internet public feelings.
Accompanying drawing explanation
Fig. 1 is present invention emotional semantic classification based on support vector machine system architecture diagram.
Detailed description of the invention
As it is shown in figure 1, as one embodiment of the present of invention, emotional semantic classification system based on support vector machine includes:
Data acquisition and pretreatment module, be responsible for utilizing web crawlers to carry out data and crawl, and what acquisition user was delivered comments
Opinion information, and review information is carried out pretreatment；
Feature Words and training sample generation module, be responsible for using the comment text through pretreatment as input, choose with
The high frequency words of specific part of speech is as Feature Words, and adds feature dictionary；Choose the evaluation text comprising Feature Words as training sample
This, and the emotion of training sample is manually marked；
Svm classifier module, is responsible for based on feature dictionary, and training sample extracts characteristic vector, and vector is supported in input
Machine generates disaggregated model；Utilize disaggregated model that the to be sorted emotion value evaluating text is calculated, analyze the emotion of text
Orientation；
Visualization model, is responsible for representing analysis result in web terminal.
Below each functional module is implemented and illustrates:
(1) data acquisition and pretreatment module (Data Collection and Preprocessing Module, letter
Claim CPM)
The main flow of data acquisition is as follows:
(1) from the beginning of the website (initial website) specified, webpage is crawled with the pattern of breadth-first；
(2) webpage got for each, resolves its page source code, letter relevant in obtaining webpage
Breath, such as: user comment information etc.；
(3) data base is write data into.
The main flow of data prediction is the Chinese word segmentation tool kit utilizing the Chinese Academy of Sciences the to research and develop evaluation text to user
Carry out participle, and mark part of speech.
(2) Feature Words and training sample generation module (Training Data Generation Module is called for short TGM)
In view of the present invention will use support vector machine (Support Vector Machine, hereinafter referred to as SVM) to comment
Text is classified, and therefore extracts one group of representative Feature Words, and chooses high-quality training sample on this basis
It is to ensure that the key of classification quality.To this end, we adopt carries out selecting of Feature Words and training sample with the following method.Main step
Rapid as follows:
(A) the choosing of Feature Words
TGM uses algorithm FindCover to choose typical Feature Words.Additionally, according to actual observation, TGM chooses part of speech
For the word of noun (n), verb (v) and adjective (a) as Feature Words, i.e. the input P of FindCover algorithm be array n,
v,a}；In this external Practical Calculation, TGM chooses the word of length L ＞ 1 as Feature Words.It is noted that for n, P and L
Value, can be adjusted according to actual needs.
Algorithm FindCover
Input: participle mark the evaluation text collection U of part of speech, Feature Words number n, Feature Words length L, part of speech set
P
Output: feature phrase
1. initialize set S, A；Here set S is the set for storage feature phrase；Here set A is for evaluating
The subset of text collection U, is specifically designed to the text id corresponding to Feature Words word deposited in S.
2. calculate mapping relations Map M, each word word is mapped to one group of text id:M comprising this word
(word)；
3. when S does not comprises n word, then find word word so that it is satisfied three conditions:
I () part of speech meets the requirement of P；
(ii) length meets the requirement of L；
(iii) current coverage rate coverage=| M (word)-A | is maximum；
4. if coverage rate coverage=0 of the word searched out, then terminate circulation, otherwise, word is added S, by M
(word) add A, return step 3 and continue cycling through, until set S comprises n word or the coverage rate of word searched out
Coverage=0；
5. return set S as feature phrase.
(B) the choosing of training sample
The feature phrase S, TGM returned according to FindCover uses following strategy to choose training sample: first, exports institute
There is the evaluation text collection U comprising Feature Wordsf.If | Uf| > 1% | U |, then from UfIn randomly choose 1% | U | individual evaluation text make
For training sample；Otherwise export UfAs training sample.Selected training sample will carry out artificial emotion mark.Actually used mistake
Cheng Zhong, can be divided into 2 classes by text according to emotion, it may be assumed that front, negatively；Also three classes, i.e. front it are divided into, neutral, negatively.
(3) svm classifier module (SVM Training Module is called for short STM)
First text in sample data is converted to shape such as according to Feature Words by STM: "<labelling>feature 1: number feature 2:
Number ... feature n: number " form, wherein according to three way classification, then<labelling>can with value as positive,
Negative or neutral；According to two way classification, then<labelling>can be with value as positive with negative.STM subsequently will
The training data changed is input in LIBSVM storehouse carry out classification based training.After obtaining training result, STM applies these to classify
Text to be sorted is calculated by rule, analyzes the orientation of emotion of text.
(4) visualization model (Visualization Module is called for short VM)
Analysis result is represented by VM at Web end, and main content viewable includes: (1) text based on particular keywords
" front ", " negatively ", the ratio of " neutral "；(2) urtext that emotion is relevant；(3) temporally dimension represents the feelings of text
Sense change.
Claims (10)
1. an emotional semantic classification system based on support vector machine, it is characterised in that including:
Data acquisition and pretreatment module, be responsible for utilizing web crawlers to carry out data and crawl, and obtains the comment letter that user is delivered
Breath, and review information is carried out pretreatment；
Feature Words and training sample generation module, be responsible for using the comment text through pretreatment as input, choose with specific
The high frequency words of part of speech is as Feature Words, and adds feature dictionary；Choose the evaluation text comprising Feature Words as training sample, and
The emotion of training sample is manually marked；
Svm classifier module, is responsible for based on feature dictionary, and training sample extracts characteristic vector, and input support vector machine is raw
Constituent class model；Utilize disaggregated model that the to be sorted emotion value evaluating text is calculated, analyze the orientation of emotion of text；
Visualization model, is responsible for representing analysis result in web terminal.
2. a sensibility classification method based on support vector machine, it is characterised in that comprise the following steps:
A, utilize web crawlers to carry out data to crawl, obtain the review information that user is delivered, and review information is carried out pre-place
Reason；
B, using the comment text through pretreatment as input, choose the high frequency words with specific part of speech as Feature Words, and add
Enter feature dictionary；Choose the evaluation text comprising Feature Words as training sample, and the emotion of training sample is manually marked
Note；
C, based on feature dictionary, to training sample extract characteristic vector, input support vector machine generate disaggregated model；Utilize
The to be sorted emotion value evaluating text is calculated by disaggregated model, analyzes the orientation of emotion of text；
D, analysis result is represented in web terminal.
A kind of sensibility classification method based on support vector machine, it is characterised in that in step A, institute
State and utilize web crawlers to carry out data to crawl, obtain the review information that user is delivered, specifically include:
From the beginning of the website specified, crawling webpage with the pattern of breadth-first, the webpage got for each, to its page
Source code resolves, and obtains user comment information in webpage, the review information write into Databasce that will obtain.
A kind of sensibility classification method based on support vector machine, it is characterised in that in step A, institute
State and review information carried out pretreatment, specifically include:
Use Chinese word segmentation tool kit that the evaluation information of user is carried out participle, and mark part of speech.
A kind of sensibility classification method based on support vector machine, it is characterised in that in step B, institute
State and choose the high frequency words with specific part of speech as Feature Words, specifically include:
It is that noun, verb and adjectival frequent words are as Feature Words based on FindCover algorithm picks part of speech.
A kind of sensibility classification method based on support vector machine, it is characterised in that described based on
FindCover algorithm picks part of speech be noun, verb and adjectival frequent words as Feature Words, method particularly includes:
Determine the input of FindCover algorithm: participle also marks the evaluation text collection U of part of speech, Feature Words number n, Feature Words
Length L, part of speech set P；
Determine the output of FindCover algorithm: feature phrase S；
The process of choosing includes:
Step 1, initialization set S, A；
Step 2, calculating mapping relations Map M, be mapped to one group of text id:M comprising this word by each word word
(word)；
Step 3, when gathering S and not comprising n word, then find word word so that it is satisfied three conditions:
I () part of speech meets the requirement of P；
(ii) length meets the requirement of L；
(iii) current coverage rate coverage=| M (word)-A | is maximum；
If coverage rate coverage=0 of the word that step 4 searches out, then terminate circulation, otherwise, word is added S, by M
(word) add A, return step 3 and continue cycling through, until set S comprises n word or the coverage rate of word searched out
Coverage=0；
Step 5, return set S are as feature phrase.
A kind of sensibility classification method based on support vector machine, it is characterised in that described n, P, L
Value can be adjusted according to practical situation.
A kind of sensibility classification method based on support vector machine, it is characterised in that in step B, institute
State and choose the evaluation text comprising Feature Words as training sample, specifically include:
The feature phrase S returned according to FindCover algorithm, uses following strategy to choose training sample: first, exports all bags
Evaluation text collection U containing Feature WordsfIf, | Uf| > 1% | U |, then from UfIn randomly choose 1% | U | individual evaluation text as instruction
Practice sample；Otherwise export UfAs training sample.
A kind of sensibility classification method based on support vector machine, it is characterised in that in step C, institute
Stating and training sample extracts characteristic vector, input support vector machine generates disaggregated model, specifically includes:
First according to Feature Words the text in sample data is converted to shape as "<labelling>feature 1: number feature 2: number ...
Feature n: number " form, according to three way classification, then<labelling>value is positive, negative or neutral；If adopting
With two way classification, then<labelling>value is positive and negative；Subsequently the training data changed is input to LIBSVM
Storehouse carries out classification based training.
A kind of sensibility classification method based on support vector machine, it is characterised in that in step D,
Analysis result is represented in web terminal, described in the content that represents include: " front " of text based on particular keywords, " negative
Face ", the ratio of " neutral ", urtext that emotion is relevant, temporally dimension represent the emotion change of text.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610529672.0A CN106126502B (en) | 2016-07-07 | 2016-07-07 | A kind of emotional semantic classification system and method based on support vector machines |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610529672.0A CN106126502B (en) | 2016-07-07 | 2016-07-07 | A kind of emotional semantic classification system and method based on support vector machines |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106126502A true CN106126502A (en) | 2016-11-16 |
| CN106126502B CN106126502B (en) | 2018-10-30 |
Family
ID=57283438
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610529672.0A Active CN106126502B (en) | 2016-07-07 | 2016-07-07 | A kind of emotional semantic classification system and method based on support vector machines |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106126502B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106407449A (en) * | 2016-09-30 | 2017-02-15 | 四川长虹电器股份有限公司 | Emotion classification method based on support vector machine |
| CN106649890A (en) * | 2017-02-07 | 2017-05-10 | 税云网络科技服务有限公司 | Data storage method and device |
| CN106682192A (en) * | 2016-12-29 | 2017-05-17 | 北京奇虎科技有限公司 | Method and device for training answer intention classification model based on search keywords |
| CN106776557A (en) * | 2016-12-13 | 2017-05-31 | 竹间智能科技（上海）有限公司 | Affective state memory recognition methods and the device of emotional robot |
| CN107229684A (en) * | 2017-05-11 | 2017-10-03 | 合肥美的智能科技有限公司 | Statement classification method, system, electronic equipment, refrigerator and storage medium |
| CN107291902A (en) * | 2017-06-23 | 2017-10-24 | 中国人民解放军国防科学技术大学 | Automatic marking method is checked in a kind of popular contribution based on hybrid classification technology |
| CN110377727A (en) * | 2019-06-06 | 2019-10-25 | 深思考人工智能机器人科技（北京）有限公司 | A kind of multi-tag file classification method and device based on multi-task learning |
| CN110689033A (en) * | 2018-07-05 | 2020-01-14 | 第四范式（北京）技术有限公司 | Data acquisition method, device and equipment for model training and storage medium |
| CN112487266A (en) * | 2019-09-12 | 2021-03-12 | 北京国双科技有限公司 | Emotion labeling method and device, computer equipment and storage medium |
| WO2021093349A1 (en) * | 2019-11-15 | 2021-05-20 | Midea Group Co., Ltd. | System, method, and user interface for facilitating product research and development |
| CN113553422A (en) * | 2021-07-16 | 2021-10-26 | 山东建筑大学 | User preference prediction method and system based on language value convolution rule inference network |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1256931A1 (en) * | 2001-05-11 | 2002-11-13 | Sony France S.A. | Method and apparatus for voice synthesis and robot apparatus |
| CN103034626A (en) * | 2012-12-26 | 2013-04-10 | 上海交通大学 | Emotion analyzing system and method |
| CN103116644A (en) * | 2013-02-26 | 2013-05-22 | 华南理工大学 | Method for mining orientation of Web themes and supporting decisions |
| CN104731770A (en) * | 2015-03-23 | 2015-06-24 | 中国科学技术大学苏州研究院 | Chinese microblog emotion analysis method based on rules and statistical model |
| CN104965822A (en) * | 2015-07-29 | 2015-10-07 | 中南大学 | Emotion analysis method for Chinese texts based on computer information processing technology |
-
2016
- 2016-07-07 CN CN201610529672.0A patent/CN106126502B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1256931A1 (en) * | 2001-05-11 | 2002-11-13 | Sony France S.A. | Method and apparatus for voice synthesis and robot apparatus |
| CN103034626A (en) * | 2012-12-26 | 2013-04-10 | 上海交通大学 | Emotion analyzing system and method |
| CN103116644A (en) * | 2013-02-26 | 2013-05-22 | 华南理工大学 | Method for mining orientation of Web themes and supporting decisions |
| CN104731770A (en) * | 2015-03-23 | 2015-06-24 | 中国科学技术大学苏州研究院 | Chinese microblog emotion analysis method based on rules and statistical model |
| CN104965822A (en) * | 2015-07-29 | 2015-10-07 | 中南大学 | Emotion analysis method for Chinese texts based on computer information processing technology |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106407449A (en) * | 2016-09-30 | 2017-02-15 | 四川长虹电器股份有限公司 | Emotion classification method based on support vector machine |
| CN106776557A (en) * | 2016-12-13 | 2017-05-31 | 竹间智能科技（上海）有限公司 | Affective state memory recognition methods and the device of emotional robot |
| CN106776557B (en) * | 2016-12-13 | 2020-09-08 | 竹间智能科技（上海）有限公司 | Emotional state memory identification method and device of emotional robot |
| CN106682192B (en) * | 2016-12-29 | 2020-07-03 | 北京奇虎科技有限公司 | Method and device for training answer intention classification model based on search keywords |
| CN106682192A (en) * | 2016-12-29 | 2017-05-17 | 北京奇虎科技有限公司 | Method and device for training answer intention classification model based on search keywords |
| CN106649890A (en) * | 2017-02-07 | 2017-05-10 | 税云网络科技服务有限公司 | Data storage method and device |
| CN106649890B (en) * | 2017-02-07 | 2020-07-14 | 税云网络科技服务有限公司 | Data storage method and device |
| CN107229684A (en) * | 2017-05-11 | 2017-10-03 | 合肥美的智能科技有限公司 | Statement classification method, system, electronic equipment, refrigerator and storage medium |
| CN107229684B (en) * | 2017-05-11 | 2021-05-18 | 合肥美的智能科技有限公司 | Sentence classification method and system, electronic equipment, refrigerator and storage medium |
| CN107291902B (en) * | 2017-06-23 | 2020-05-08 | 中国人民解放军国防科学技术大学 | Automatic marking method for public contribution review based on mixed classification technology |
| CN107291902A (en) * | 2017-06-23 | 2017-10-24 | 中国人民解放军国防科学技术大学 | Automatic marking method is checked in a kind of popular contribution based on hybrid classification technology |
| CN110689033A (en) * | 2018-07-05 | 2020-01-14 | 第四范式（北京）技术有限公司 | Data acquisition method, device and equipment for model training and storage medium |
| CN110377727A (en) * | 2019-06-06 | 2019-10-25 | 深思考人工智能机器人科技（北京）有限公司 | A kind of multi-tag file classification method and device based on multi-task learning |
| CN110377727B (en) * | 2019-06-06 | 2022-06-17 | 深思考人工智能机器人科技（北京）有限公司 | Multi-label text classification method and device based on multi-task learning |
| CN112487266A (en) * | 2019-09-12 | 2021-03-12 | 北京国双科技有限公司 | Emotion labeling method and device, computer equipment and storage medium |
| WO2021093349A1 (en) * | 2019-11-15 | 2021-05-20 | Midea Group Co., Ltd. | System, method, and user interface for facilitating product research and development |
| CN113553422A (en) * | 2021-07-16 | 2021-10-26 | 山东建筑大学 | User preference prediction method and system based on language value convolution rule inference network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106126502B (en) | 2018-10-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106126502B (en) | A kind of emotional semantic classification system and method based on support vector machines | |
| Li et al. | Twiner: named entity recognition in targeted twitter stream | |
| CN106294593B (en) | In conjunction with the Relation extraction method of subordinate clause grade remote supervisory and semi-supervised integrated study | |
| CN103678564B (en) | Internet product research system based on data mining | |
| CN106407236B (en) | A kind of emotion tendency detection method towards comment data | |
| CN105528437B (en) | A kind of question answering system construction method extracted based on structured text knowledge | |
| CN104881458B (en) | A kind of mask method and device of Web page subject | |
| CN101127042A (en) | Sensibility classification method based on language model | |
| CN102200975B (en) | Vertical search engine system using semantic analysis | |
| CN104268160A (en) | Evaluation object extraction method based on domain dictionary and semantic roles | |
| CN106096664A (en) | A kind of sentiment analysis method based on social network data | |
| CN103246644B (en) | Method and device for processing Internet public opinion information | |
| CN105843796A (en) | Microblog emotional tendency analysis method and device | |
| CN110413787B (en) | Text clustering method, device, terminal and storage medium | |
| CN104008091A (en) | Sentiment value based web text sentiment analysis method | |
| CN108376133A (en) | The short text sensibility classification method expanded based on emotion word | |
| CN106202584A (en) | A kind of microblog emotional based on standard dictionary and semantic rule analyzes method | |
| CN105183715B (en) | A kind of word-based distribution and the comment spam automatic classification method of file characteristics | |
| CN107451118A (en) | Sentence-level sensibility classification method based on Weakly supervised deep learning | |
| CN102929861A (en) | Method and system for calculating text emotion index | |
| CN103761239A (en) | Method for performing emotional tendency classification to microblog by using emoticons | |
| CN108763348A (en) | A kind of classification improved method of extension short text word feature vector | |
| CN105512333A (en) | Product comment theme searching method based on emotional tendency | |
| CN104899335A (en) | Method for performing sentiment classification on network public sentiment of information | |
| CN106407449A (en) | Emotion classification method based on support vector machine |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |