CN104317965A - Establishment method of emotion dictionary based on linguistic data - Google Patents
Establishment method of emotion dictionary based on linguistic data Download PDFInfo
- Publication number
- CN104317965A CN104317965A CN201410649358.7A CN201410649358A CN104317965A CN 104317965 A CN104317965 A CN 104317965A CN 201410649358 A CN201410649358 A CN 201410649358A CN 104317965 A CN104317965 A CN 104317965A
- Authority
- CN
- China
- Prior art keywords
- word
- list
- polarity
- adjective
- subordinate 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.)
- Granted
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
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2462—Approximate or statistical queries
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Machine Translation (AREA)
Abstract
The invention discloses an establishment method of an emotion dictionary based on linguistic data, and the establishment method comprises the following steps of: in advance acquiring partial known emotional tendency adjectives, including positive and negative adjectives, extracting and analyzing the unknown emotional tendency adjectives by using adversatives and privatives, constantly extending seed lexicons and finally judging them. The method doesn't need manual intervention and belongs to a non-supervision learning method so as to greatly improve the operating efficiency. The emotion dictionary established by the method can be applied for critical analysis so as to quickly obtain emotional tendency, thereby achieving the purpose of rapid analysis.
Description
Technical field
The invention belongs to artificial intelligence invention technology, be specifically related to a kind of sentiment dictionary construction method based on language material.
Background technology
Existing part Chinese sentiment dictionary, be all build by artificially summing up some conventional adjectives, inefficiency, does not have territoriality again.And Chinese is not similar to the dictionary of English wordnet, emotion word dictionary cannot be built by existing dictionary.Based on the sentiment dictionary construction method of language material, the speech habits of people are applied in the analysis of text, construct actively and passive two class dictionaries.Namely save labor cost, there is again territoriality and the judgment to neologisms emotion.
Comparatively early analyze language material according to language rule, that build sentiment dictionary is Hazivassiloglou and McKeown, they utilize a corpus and adjective emotion word subset, according to language rule, find out other adjectival emotions and point to.Such as utilize " AND ", conjunctions such as " BUT ".They also use clustering algorithm to determine two words that conjunction couples together to have identical or contrary polarity, thus produce two set of words.Kanayama and Nasukawa uses in sentence and between sentence, the conforming concept of emotion generates sentiment dictionary, and between sentence, consistance is because the sentence of usual phase feeling of sympathy is ined succession.Feeling changes is normally caused by adversative, such as " but ".But they are poor to the utilization factor of the emotion word obtained in algorithm implementation.
Summary of the invention
The object of the present invention is to provide a kind of sentiment dictionary generation method based on large-scale corpus, it is high that the method has accuracy rate, and the advantage such as to save time, can be the reference that comment and analysis provides important.
The technical scheme realizing the object of the invention is: a kind of sentiment dictionary construction method based on language material, comprises the following steps:
The first step, utilizes Chinese word segmentation instrument, carries out pre-service to language material, continuous print Chinese sentence in language material is divided into word or word one by one, separates with space, and the part of speech of tagged words or word;
Second step, count all adjectival word frequency in language material and by sorting from high to low, get front 5%-10% to have and determine that the adjective of feeling polarities forms emotion dictionary as seed words, and analyze the feeling polarities of seed words, be called positive by the polarity of word that front is evaluated, it is passive the polarity of the word of unfavorable ratings to be called, forms two seed words lists respectively, these two seed words lists are as the initial list of emotion dictionary, and initial word frequency is 1;
3rd step, the text of learning from else's experience in pretreated language material, the language material analyzed if necessary, makes pauses in reading unpunctuated ancient writings to text according to punctuate, obtains multiple subordinate sentence, not containing punctuate in subordinate sentence, continues execution the 4th step; If there is no the language material of Water demand, then go to the 6th step;
4th step, search for the adjective in each subordinate sentence obtained, set a threshold k, travel through within the scope of K word or word before adjective position, according to the word with Negation pointed out in Chinese dictionary, judge whether negative word, if having, then add in respective list according to polarity transition rule, otherwise stop finding negative word; Again according to the word with turnover meaning pointed out in Chinese dictionary, judge that whether this subordinate sentence is with adversative beginning, if so, then change current polarity according to polarity transition rule, otherwise polarity is constant; Then by polarity transition rule, the adjective in subordinate sentence is added in two list s and a respectively;
5th step, analyze the polarity of two list s and a that the 4th step obtains, namely by the polarity of seed words inspection list s and a in emotion dictionary, if the number containing positive seed words in one of them list is no less than passive seed words, then in this list, all words are classified as positive, and the word in another list is then classified as passive; If the passive seed words respectively containing equal number in two lists and positive seed words, then return the 3rd step; Otherwise, judge that the adjective of polarity adds to as seed words in the initial list of emotion dictionary using in two list s and a, if this adjective existing in initial list, then its word frequency is added 1, otherwise to arrange this adjectival word frequency be 1, return the 3rd step;
6th step, travels through the final emotion dictionary obtained, to being judged as positive and passive word simultaneously, gets its word frequency, if it is high to belong to positive word frequency, then this word is positive, otherwise is passive.
The present invention compares with existing emotion word construction method, and its remarkable advantage is: (1) saves time and labor cost.The method according to language material Automatic Extraction adjective as emotion word, and can differentiate its feeling polarities, generates actively and passive two sentiment dictionaries.Greatly save time and energy than artificial mark.(2) reliability is strong.This sentiment dictionary construction method repeatedly follows majority principle, processes for some disturbed conditions that may occur, to ensure the degree of accuracy of algorithm.(3) highly versatile.This algorithm according to the language material of every field, can generate field emotion word.(4) sentiment dictionary generated can be comment and analysis, and natural language processing provides important reference frame.
Accompanying drawing explanation
Fig. 1 is that emotion word list ownership differentiates process flow diagram.
Fig. 2 is sentiment dictionary product process figure.
Embodiment
The inventive method comprises the following steps:
The first step: utilize the language material having carried out word segmentation processing, these language materials are generally subjective comments, the adjective of appearance is sorted from high to low by word frequency, extract front 5% ~ 10% and have the adjective determining feeling polarities, according to knowing that net (Hownet) marks its feeling polarities as seed words, form emotion dictionary;
Second step: by punctuation mark punctuate, generate short sentence one by one;
3rd step: short sentence scanning one by one, extracts adjective.Build two temporary table, for storing the adjective of two feeling polarities.For adjective, judge whether it has negative word to modify, if had, polarity negate, stored in corresponding temporary table; Judge that whether short sentence is with adversative beginning again, if so, polarity is negate again, stored in corresponding temporary table;
4th step: for the temporary table obtained, goes to judge with existing seed dictionary.The list many containing positive emotion word is then the list of positive emotion word, otherwise is the list of Negative Affect word.Here existing emotion word is utilized to go to judge unknown emotion word, according to majority principle, because the word in list all has phase feeling of sympathy, if the word list of a unknown feeling polarities contains the seed words of more positive emotion, so just think that the emotion word in list is all positive, if instead the seed words containing more Negative Affect, then all words in list think it is all passive, and this can suitably reduce the error caused because language material is lack of standardization indirectly;
5th step: the new emotion word that the 4th step obtains is added in seed words, as the foundation that next time judges.The original emotion dictionary of so continuous expansion, make along with reading increasing of language material, the accuracy rate of judgement improves constantly.
6th step: after having read all language materials, obtains two last emotion word dictionaries.Wherein may there be some words, namely in the dictionary of positive emotion, again in the dictionary of Negative Affect.Because language material may exist lack of standard, word misjudgment may be caused, now according to majority principle, according to the in most cases generic judgement of this word.
Below in conjunction with accompanying drawing, the present invention is described in further detail.
The present invention is based on the sentiment dictionary building method of language material, first utilize existing participle instrument, participle and part-of-speech tagging process are carried out to original language material.Then extract word frequency the highest 5% ~ 10%, and according to the sentiment analysis word collection knowing that net (Hownet) provides, mark feeling polarities, as seed words.Make pauses in reading unpunctuated ancient writings according to punctuate then to language material, then scan corpus, for each comment, adjective is divided into opposite polarity two classes, then judges with existing seed words.Obtain its feeling polarities, they are continued to judge comment below as seed words.Finally eliminate the situation simultaneously belonging to two classes, obtain final sentiment dictionary.All applicable under any field of the method, only need provide corresponding language material.
Fig. 1 is the generative process of adjective list, can find out, for an adjective, needs according to or without negative word, and with or without adversative, integral polarity three conditions that whether change judge.Composition graphs 2, the present invention is based on the sentiment dictionary construction algorithm of language material, step is as follows:
The first step: carry out participle, part-of-speech tagging process to original language material, language material is generally the comment statement with subjective colo(u)r.Here can use some Open-Source Tools, such as: knot, ICTCLAS etc., the word separated one by one or word and its part of speech, punctuate is also taken as a word processing.
Second step: scan all comments, obtain all adjective word frequency, get word frequency the highest front 5% ~ 10%, utilization is known " word (Chinese) is evaluated in front " and " unfavorable ratings word (Chinese) " in ' sentiment analysis word collection ' that net (Hownet) provides, mark its feeling polarities, as seed words, form initial emotion dictionary.
3rd step: the text of learning from else's experience in pretreated language material, if having, makes pauses in reading unpunctuated ancient writings to text according to punctuate, obtains multiple subordinate sentence, not containing punctuate in subordinate sentence; If no, then go to the 7th step;
4th step: to the comment containing many words, scan sentence by sentence, find adjective, namely suffix is the/word of a or/an, candidate's emotion word that Here it is.
5th step: for the complete sentence of in language material, three variablees are set.Current polarity plor, is initially 1; Deposit two list s and a of emotion word, deposit the emotion word (but its concrete emotion unknown) of two polarity respectively, deposit the word identical with initial polarity in s, in a, deposit the word contrary with initial emotion.Search and locate an adjective, arrange window value size, desirable different value, is generally 3 as required, judges whether have negative word on the left side window ranges, if having, then adds in respective list according to polarity transition rule, otherwise stops finding negative word; Again according to the word with turnover meaning pointed out in Chinese dictionary, judge whether this subordinate sentence starts with adversative, if then change current polarity according to polarity transition rule, namely Plor gets negative, otherwise polarity is constant, then by polarity transition rule, the adjective in subordinate sentence is added in two list s and a respectively.It is as follows that emotion word puts into list (ACL) regulations:
A. have adversative not have negative word, if plor=1, illustrate that polarity upset does not occur or overturns identical with original polarity through even-times, then adjective is put in list a, Plor gets negative; Otherwise if plor=-1, illustrate that polarity generation odd-times overturns, then put into s, Plor gets negative.
B. have adversative to have negative word, if plor=1, adjective is put in s, and Plor gets negative; Otherwise put into a, Plor gets negative.
C. do not have adversative to have negative word, if plor=1, adjective is put in list a, and Plor gets negative; Otherwise put into s, Plor gets negative.
D. do not have adversative not have negative word, if plor=1, adjective is put in list s, and Plor gets negative; Otherwise put into a, Plor gets negative.
Give an example: " overall/n is pretty good/a./ wp is exactly/d a little/d is expensive/a, logistics/n not /d is /v very/d is fast/a.But/d still/d very/d satisfaction/a /u once/q net purchase/n ,/wp praises/v! / wp " initial plor=1.Find adjective after punctuate, first find " well ", plor=1, look for negative word according to window size, do not find, beginning of the sentence is not adversative, and " well " is put into s list; Find again " expensive ", before not negative, have " being exactly " adversative, so " expensive " and adjective reversal of poles before, because plor=1, so put into a list, while plor become-1; 3rd adjective " soon ", has negative word, does not have adversative, because plor=-1, so according to above-mentioned rule, puts into s list; Last adjective " satisfaction ", negative, does not have turnover, now because plor=-1, puts into s list, and plor becomes 1. end products simultaneously: be " well " in s, " soon ", " satisfaction "; In a be " expensive ".
6th step: all adjectives in commenting on are regularly divided in two lists, go with seed words the polarity judging word in these two lists.If the seed words number containing positive emotion in a list is more than the seed words number of Negative Affect, then the seed words list of positive emotion all put in all in this list adjectives, and another then puts into the seed words list of Negative Affect.
7th step: finally obtain two sentiment dictionaries and the word frequency of emotion word in language material.Because the lack of standard of language material, may there is certain word and occur in two dictionaries simultaneously, this is not allowed to.So if this word belongs to the word frequency that positive word frequency is greater than the passiveness after weighting, then the emotion of this word is positive, otherwise be considered as passive.
The inventive method makes full use of the emotion word obtained in algorithmic procedure, constantly expands seed words.Finally by comparing, according to majority principle, differentiate emotion, energy exclusive segment interfering data, accuracy rate is relatively high.
Claims (2)
1., based on a sentiment dictionary construction method for language material, it is characterized in that comprising the following steps:
The first step, utilizes Chinese word segmentation instrument, carries out pre-service to language material, continuous print Chinese sentence in language material is divided into word or word one by one, separates with space, and the part of speech of tagged words or word;
Second step, count all adjectival word frequency in language material and by sorting from high to low, get front 5%-10% to have and determine that the adjective of feeling polarities forms emotion dictionary as seed words, and analyze the feeling polarities of seed words, be called positive by the polarity of word that front is evaluated, it is passive the polarity of the word of unfavorable ratings to be called, forms two seed words lists respectively, these two seed words lists are as the initial list of emotion dictionary, and initial word frequency is 1;
3rd step, the text of learning from else's experience in pretreated language material, the language material analyzed if necessary, makes pauses in reading unpunctuated ancient writings to text according to punctuate, obtains multiple subordinate sentence, not containing punctuate in subordinate sentence, continues execution the 4th step; If there is no the language material of Water demand, then go to the 6th step;
4th step, search for the adjective in each subordinate sentence obtained, set a threshold k, travel through within the scope of K word or word before adjective position, according to the word with Negation pointed out in Chinese dictionary, judge whether negative word, if having, then add in respective list according to polarity transition rule, otherwise stop finding negative word; Again according to the word with turnover meaning pointed out in Chinese dictionary, judge that whether this subordinate sentence is with adversative beginning, if so, then change current polarity according to polarity transition rule, otherwise polarity is constant; Then by polarity transition rule, the adjective in subordinate sentence is added in two list s and a respectively;
5th step, analyze the polarity of two list s and a that the 4th step obtains, namely by the polarity of seed words inspection list s and a in emotion dictionary, if the number containing positive seed words in one of them list is no less than passive seed words, then in this list, all words are classified as positive, and the word in another list is then classified as passive; If the passive seed words respectively containing equal number in two lists and positive seed words, then return the 3rd step; Otherwise, judge that the adjective of polarity adds to as seed words in the initial list of emotion dictionary using in two list s and a, if this adjective existing in initial list, then its word frequency is added 1, otherwise to arrange this adjectival word frequency be 1, return the 3rd step;
6th step, travels through the final emotion dictionary obtained, to being judged as positive and passive word simultaneously, gets its word frequency, if it is high to belong to positive word frequency, then this word is positive, otherwise is passive.
2. the sentiment dictionary construction method based on language material according to claim 1, is characterized in that the polarity transition rule in described 4th step is specific as follows:
Plor variable is set, for representing whether the polarity between subordinate sentence shifts, and is initially 1, there is even-times adversative in 1 expression, the subordinate sentence that punctuate is connected is identical with initial polarity, and-1 expression occurs that odd-times is transferred, and the subordinate sentence that punctuate is connected is contrary with initial polarity; Acquiescence is put in the list s identical with initial polarity, puts the word contrary with initial polarity, be divided into four kinds of situations in list a:
Subordinate sentence beginning has adversative, does not have negative word, if plor=1, adjective is put in list a, otherwise puts into list s before adjective in K word or word, and Plor becomes-1;
Subordinate sentence beginning has adversative, has negative word, if plor=1, adjective is put in s, otherwise put into list a before adjective in K word or word, and Plor becomes-1;
Subordinate sentence beginning does not have adversative, has negative word, if plor=1, adjective is put in list a, otherwise puts into list s before adjective in K word or word;
Subordinate sentence beginning does not have adversative, does not have negative word, if plor=1, adjective is put in list s, otherwise puts into list a before adjective in K word or word.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410649358.7A CN104317965B (en) | 2014-11-14 | 2014-11-14 | Sentiment dictionary construction method based on language material |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410649358.7A CN104317965B (en) | 2014-11-14 | 2014-11-14 | Sentiment dictionary construction method based on language material |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104317965A true CN104317965A (en) | 2015-01-28 |
| CN104317965B CN104317965B (en) | 2018-04-03 |
Family
ID=52373197
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410649358.7A Expired - Fee Related CN104317965B (en) | 2014-11-14 | 2014-11-14 | Sentiment dictionary construction method based on language material |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104317965B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105095190A (en) * | 2015-08-25 | 2015-11-25 | 众联数据技术（南京）有限公司 | Chinese semantic structure and finely segmented word bank combination based emotional analysis method |
| CN105760502A (en) * | 2016-02-23 | 2016-07-13 | 常州普适信息科技有限公司 | Commercial quality emotional dictionary construction system based on big data text mining |
| CN106610955A (en) * | 2016-12-13 | 2017-05-03 | 成都数联铭品科技有限公司 | Dictionary-based multi-dimensional emotion analysis method |
| CN107608953A (en) * | 2017-07-25 | 2018-01-19 | 同济大学 | A kind of term vector generation method based on random length context |
| CN108563635A (en) * | 2018-04-04 | 2018-09-21 | 北京理工大学 | A kind of sentiment dictionary fast construction method based on emotion wheel model |
| CN108647191A (en) * | 2018-05-17 | 2018-10-12 | 南京大学 | It is a kind of based on have supervision emotion text and term vector sentiment dictionary construction method |
| CN108694165A (en) * | 2017-04-10 | 2018-10-23 | 南京理工大学 | Cross-cutting antithesis sentiment analysis method towards product review |
| CN109800418A (en) * | 2018-12-17 | 2019-05-24 | 北京百度网讯科技有限公司 | Text handling method, device and storage medium |
| CN110287319A (en) * | 2019-06-13 | 2019-09-27 | 南京航空航天大学 | Students' evaluation text analyzing method based on sentiment analysis technology |
| CN112905736A (en) * | 2021-01-27 | 2021-06-04 | 郑州轻工业大学 | Unsupervised text emotion analysis method based on quantum theory |
| CN115796158A (en) * | 2023-02-07 | 2023-03-14 | 中国传媒大学 | Emotion dictionary construction method and device, electronic equipment and computer readable medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102323944A (en) * | 2011-09-02 | 2012-01-18 | 苏州大学 | Sentiment classification method based on polarity transfer rules |
| CN103020249A (en) * | 2012-12-19 | 2013-04-03 | 苏州大学 | Classifier construction method and device as well as Chinese text sentiment classification method and system |
-
2014
- 2014-11-14 CN CN201410649358.7A patent/CN104317965B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102323944A (en) * | 2011-09-02 | 2012-01-18 | 苏州大学 | Sentiment classification method based on polarity transfer rules |
| CN103020249A (en) * | 2012-12-19 | 2013-04-03 | 苏州大学 | Classifier construction method and device as well as Chinese text sentiment classification method and system |
Non-Patent Citations (1)
| Title |
|---|
| HATZIVASSILOGLOU V, MCKEOWN K R: "Predicting the Semantic Orientation of Adjectives", 《PROCEEDINGS OF THE 35TH ANNUAL MEETING OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS AND EIGHTH CONFERENCE OF THE EUROPEAN CHAPTER OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS. STROUDSBURG, PA, USA: ASSOCIATION FOR COMPUTATIONAL LINGUISTICS》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105095190B (en) * | 2015-08-25 | 2018-01-12 | 众联数据技术（南京）有限公司 | A kind of sentiment analysis method combined based on Chinese semantic structure and subdivision dictionary |
| CN105095190A (en) * | 2015-08-25 | 2015-11-25 | 众联数据技术（南京）有限公司 | Chinese semantic structure and finely segmented word bank combination based emotional analysis method |
| CN105760502A (en) * | 2016-02-23 | 2016-07-13 | 常州普适信息科技有限公司 | Commercial quality emotional dictionary construction system based on big data text mining |
| CN106610955A (en) * | 2016-12-13 | 2017-05-03 | 成都数联铭品科技有限公司 | Dictionary-based multi-dimensional emotion analysis method |
| CN108694165A (en) * | 2017-04-10 | 2018-10-23 | 南京理工大学 | Cross-cutting antithesis sentiment analysis method towards product review |
| CN108694165B (en) * | 2017-04-10 | 2021-11-09 | 南京理工大学 | Cross-domain dual emotion analysis method for product comments |
| CN107608953A (en) * | 2017-07-25 | 2018-01-19 | 同济大学 | A kind of term vector generation method based on random length context |
| CN108563635A (en) * | 2018-04-04 | 2018-09-21 | 北京理工大学 | A kind of sentiment dictionary fast construction method based on emotion wheel model |
| CN108647191A (en) * | 2018-05-17 | 2018-10-12 | 南京大学 | It is a kind of based on have supervision emotion text and term vector sentiment dictionary construction method |
| CN109800418A (en) * | 2018-12-17 | 2019-05-24 | 北京百度网讯科技有限公司 | Text handling method, device and storage medium |
| CN109800418B (en) * | 2018-12-17 | 2023-05-05 | 北京百度网讯科技有限公司 | Text processing method, device and storage medium |
| CN110287319A (en) * | 2019-06-13 | 2019-09-27 | 南京航空航天大学 | Students' evaluation text analyzing method based on sentiment analysis technology |
| CN110287319B (en) * | 2019-06-13 | 2021-06-15 | 南京航空航天大学 | Student evaluation text analysis method based on emotion analysis technology |
| CN112905736A (en) * | 2021-01-27 | 2021-06-04 | 郑州轻工业大学 | Unsupervised text emotion analysis method based on quantum theory |
| CN112905736B (en) * | 2021-01-27 | 2023-09-19 | 郑州轻工业大学 | Quantum theory-based unsupervised text emotion analysis method |
| CN115796158A (en) * | 2023-02-07 | 2023-03-14 | 中国传媒大学 | Emotion dictionary construction method and device, electronic equipment and computer readable medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104317965B (en) | 2018-04-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104317965A (en) | Establishment method of emotion dictionary based on linguistic data | |
| CN107092596B (en) | Text emotion analysis method based on attention CNNs and CCR | |
| CN110717339B (en) | Semantic representation model processing method and device, electronic equipment and storage medium | |
| Ghosh et al. | Fracking sarcasm using neural network | |
| CN106407236B (en) | A kind of emotion tendency detection method towards comment data | |
| CN107480143B (en) | Method and system for segmenting conversation topics based on context correlation | |
| Mishra et al. | Sentiment analysis of Twitter data: Case study on digital India | |
| CN106649603B (en) | Designated information pushing method based on emotion classification of webpage text data | |
| CN104268160A (en) | Evaluation object extraction method based on domain dictionary and semantic roles | |
| CN105183717B (en) | A kind of OSN user feeling analysis methods based on random forest and customer relationship | |
| CN104881402A (en) | Method and device for analyzing semantic orientation of Chinese network topic comment text | |
| CN102200969A (en) | Text sentiment polarity classification system and method based on sentence sequence | |
| CN102096680A (en) | Method and device for analyzing information validity | |
| CN104346326A (en) | Method and device for determining emotional characteristics of emotional texts | |
| Stavrianou et al. | NLP-based feature extraction for automated tweet classification | |
| CN110321434A (en) | A kind of file classification method based on word sense disambiguation convolutional neural networks | |
| CN104778240A (en) | Micro blog text data classification method on basis of multi-feature fusion | |
| CN105224520A (en) | A kind of Chinese patent documentation term automatic identifying method | |
| Samsudin et al. | Mining opinion in online messages | |
| CN107341142B (en) | Enterprise relation calculation method and system based on keyword extraction and analysis | |
| CN104572632A (en) | Method for determining translation direction of word with proper noun translation | |
| CN114065749A (en) | Text-oriented Guangdong language recognition model and training and recognition method of system | |
| CN102591976A (en) | Text characteristic extracting method and document copy detection system based on sentence level | |
| Baniata et al. | Sentence representation network for Arabic sentiment analysis | |
| CN103942188B (en) | A kind of method and apparatus identifying language material language |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180403 Termination date: 20201114 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |