CN105068988B - Various dimensions and more granularity sentiment analysis methods - Google Patents
Various dimensions and more granularity sentiment analysis methods Download PDFInfo
- Publication number
- CN105068988B CN105068988B CN201510431992.8A CN201510431992A CN105068988B CN 105068988 B CN105068988 B CN 105068988B CN 201510431992 A CN201510431992 A CN 201510431992A CN 105068988 B CN105068988 B CN 105068988B
- Authority
- CN
- China
- Prior art keywords
- information
- sentiment
- resources
- sentiment analysis
- social
- 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.)
- Active
Links
- 238000004458 analytical method Methods 0.000 title claims abstract description 119
- 230000008451 emotion Effects 0.000 claims abstract description 47
- 235000019580 granularity Nutrition 0.000 claims abstract description 27
- 230000000694 effects Effects 0.000 claims abstract description 4
- 238000000034 method Methods 0.000 claims description 19
- 230000008569 process Effects 0.000 claims description 10
- 239000002245 particle Substances 0.000 claims description 5
- 239000004744 fabric Substances 0.000 claims description 3
- 238000012512 characterization method Methods 0.000 claims 2
- 238000010276 construction Methods 0.000 abstract description 3
- 238000012545 processing Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000002996 emotional effect Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000007935 neutral effect Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention relates to a kind of various dimensions and more granularity sentiment analysis methods, including:Affection resources are built, i.e., according to its affection resources of the classification system construction of specific area text;Sentiment orientation word is selected, that is, selects the emotion word under each classification and determines its Sentiment orientation;Differentiate emotion tendency, including:Judge the type of information resources;Emotion keyword is obtained from information resources;Authoritative publisher is identified from information resources, and obtains the sentiment analysis result of the information resources;Sentiment analysis is carried out to social category information;The Sentiment orientation of non-proprietary classification social activity category information is analyzed;Sentiment analysis is carried out for the social information of proprietary classification.The sentiment analysis method of the present invention can carry out sentiment analysis to provide higher sentiment analysis discrimination and precision from various dimensions, more granularities.
Description
Technical field
The present invention relates to semantic analysis technology field, relates more specifically to a kind of various dimensions and more granularity sentiment analysis sides
Method.
Background technology
With the rapid development of Internet, the information contained in internet occupies more in daily life
Carry out more consequence.The especially rise of mobile Internet so that people can be in the mobile terminal quick obtaining net such as mobile phone
Network information, this has also further speeded up the development of the internet such as network social intercourse media information source.Sentiment analysis be to
The process that the text of emotional color or Sentiment orientation is handled, analyzed and applied, and in natural language processing field compared with
For the research direction in forward position.With the explosive increase of the network information, receive much concern public sentiment monitoring, event prediction, commodity
The fields such as recommendation, the importance more and more higher that the emotional color contained to information is automatically analyzed.Therefore, for network,
The research tool that the information of the specific areas such as SMS carries out sentiment analysis is of great significance.
At present, the research on sentiment analysis mainly has the method based on sentiment dictionary and the method based on machine learning.
Wherein the method based on sentiment dictionary relies primarily on general sentiment dictionary, by count occur in text it is legal short
The semantic tendency of language, to judge the Sentiment orientation of whole text.The shortcomings that this method be can not towards specific area, such as
There is positive emotion tendency in " underground organization " one word, but but carried in the field such as anti-corruption in the field text for eulogizing Red Army
Negative emotion is inclined to.Although the method based on machine learning can not be constrained by dictionary, due to needing largely have just
The training corpus really marked, and when target text and training sample deviation it is larger, especially change the short texts such as various microblogging
When, it can not often obtain preferable sentiment analysis effect.
The defects of current sentiment analysis method, is mainly shown as described below:1st, discrimination is low, due to nets such as microbloggings
The lack of standard of network text so that sentiment analysis discrimination of traditional method in this section in data is relatively low;2nd, analysis precision
It is not high, because same word may show as different Sentiment orientations in different classes of text, when suitable for some classification
Affection resources when being applied to the text of another classification, the availability of method can be caused to decline, it is therefore desirable to improve analysis
Precision, corresponding affection resources are respectively adopted for the text under different classes of in field and are analyzed;3rd, dimension is analyzed not
Foot, the information resources on internet, especially microblogging etc. often have multidimensional property, if only from information contained text in itself
This single dimension of content is analyzed its Sentiment orientation, tends to lose the useful information contained by resource itself, because
This needs the information for considering multiple dimensions contained by information resources itself, is believed with farthest excavating the emotion contained in resource
Breath.
The content of the invention
In view of the above-mentioned problems in the prior art, it is a primary object of the present invention to solve the defects of prior art,
And invent and a kind of can carry out sentiment analysis from various dimensions, more granularities to provide the feelings of higher sentiment analysis discrimination and precision
Feel analysis method.
The invention provides a kind of various dimensions and more granularity sentiment analysis methods, comprise the following steps:
Step S1, affection resources are built, i.e., according to the real needs of sentiment analysis and the class complicated variant of domain-oriented text
System, build the affection resources to be matched with the domain class complicated variant system;
Step S2, Sentiment orientation word is selected, i.e., according to the real needs of sentiment analysis, select each class in institute's domain-oriented
Emotion word under not simultaneously determines its Sentiment orientation;
Step S3, differentiate emotion tendency, comprise the following steps:According to the characteristic of the information resources of Sentiment orientation to be discriminated
Judge the type belonging to it;Emotion keyword in the affection resources, provided from the information of the Sentiment orientation to be discriminated
Emotion keyword is obtained in source;Authoritative publisher is identified from the information resources of the Sentiment orientation to be discriminated, and according to described
The polarity type of authoritative publisher obtains the sentiment analysis result belonging to described information resource;Emotion point is carried out to social category information
Analysis;To event category for general categories information or can not obtain sentiment analysis result social category information carry out Sentiment orientation
Analysis;For event category in social information sentiment analysis is carried out for the information of proprietary classification.
Wherein, the affection resources are various dimensions, the resource hierarchy of more granularities, and wherein various dimensions are used to characterize taxonomic hierarchies
In multiple class categories, more granularities are used to characterize the size of used characteristic particle size under certain dimension, for institute towards neck
Each class categories in domain, according to the real needs of sentiment analysis can symbolization, word, phrase and sentence etc. be no simultaneously
The character string of one-size and its corresponding Sentiment orientation in the category are as feature.
Wherein, the selection standard of the Sentiment orientation word is:It is selected in same granularity level under different classes of dimension
Emotion keyword it is incomplete same;Meanwhile for same emotion keyword, can have not under different classification dimensions
Same Sentiment orientation, its specific Sentiment orientation should be selected according to the specific category dimension belonging to it.
Wherein, the sentiment analysis method also includes constructing general affection resources and proprietary affection resources.
Wherein, the proprietary affection resources include authoritative account affection resources, social information viewpoint sentence affection resources, social activity
Information phrase affection resources, social information emoticon affection resources and social information word affection resources.
Wherein, in the step of carrying out sentiment analysis to social category information described in step S3, according to the emotion to be discriminated
Whether the publisher of the information resources of tendency, which is authoritative publisher, whether type is social information, whether content includes is commented on, letter
Whether breath is the multimedia messages such as video, whether property is news, whether affiliated event is general categories and the information resources
The class categories attribute in totally seven dimensions belonging in classification system, sentences to the emotion tendency of information resources respectively
It is disconnected.
Wherein, in the step of carrying out sentiment analysis to social category information described in step S3, according to the authoritative account constructed
Number affection resources judge the polarity of emotion, and social information sentiment analysis or use are being called for the information of authoritative account issue
The Sentiment orientation of information can be rapidly and accurately judged before general information sentiment analysis step.
Wherein, event category in social information is directed to described in step S3 and carries out sentiment analysis for the information of proprietary classification
Step includes:
Before formally proprietary dictionary analysis process is entered, the length of the core content text of described information is judged, if long
It is neutral to spend for 0 Sentiment orientation for thinking described information, otherwise into the proprietary dictionary analysis process;
The viewpoint sentence in described information is extracted, if viewpoint sentence be present, institute is counted according to proprietary viewpoint sentence affection resources
The viewpoint sentence quantity hit in information is stated, according to the feeling polarities of the viewpoint sentence set analysis described information of hit;
The phrase in described information is extracted, if phrase be present, according in proprietary phrase affection resources statistics described information
The phrase quantity of hit, according to the feeling polarities of the phrase set analysis described information of hit;
The emoticon in described information is extracted, if emoticon be present, is united according to proprietary emoticon affection resources
The emoticon quantity hit in meter described information, according to the feeling polarities of the emoticon set analysis described information of hit;
The word in described information is extracted, if word be present, according in proprietary word affection resources statistics described information
The word quantity of hit, the feeling polarities of described information are analyzed according to the set of words of hit, otherwise it is assumed that the feelings of described information
Sense polarity is unidentified polarity.
Understand that sentiment analysis method of the invention can be from the hair of various dimensions, i.e. information resources based on above-mentioned technical proposal
Whether cloth person is authoritative publisher, whether type is social information, whether content be more matchmakers such as video comprising comment, information
Whether body information, property are news, whether affiliated event is that general categories and the information resources are affiliated in classification system
Totally seven dimensions, and more granularities such as class categories, i.e., the varigrained character string such as symbol, word, phrase and sentence and
The angle of the Sentiment orientation in the category corresponding to it carries out sentiment analysis, knows so as to provide higher sentiment analysis
Not rate and precision.
Brief description of the drawings
Fig. 1 is the overall flow block diagram of the sentiment analysis method of the present invention;
Fig. 2 is the main flow block diagram of the sentiment analysis processing of the present invention;
Fig. 3 is the processing rule and flow diagram that social affective is analyzed in the present invention;
Fig. 4 is the processing rule and flow diagram for carrying out sentiment analysis in the present invention using proprietary dictionary.
Embodiment
For the object, technical solutions and advantages of the present invention are more clearly understood, below in conjunction with specific embodiment, and reference
Accompanying drawing, the present invention is described in further detail.
The invention discloses a kind of various dimensions and more granularity sentiment analysis methods, comprise the following steps:
Step S1, affection resources system is built, is specifically included:According to the real needs of sentiment analysis and domain-oriented
The classification system of text, build the affection resources to be matched with the domain class complicated variant system;
Step S2, Sentiment orientation word is selected, is specifically included:According to the real needs of sentiment analysis, institute's domain-oriented is selected
In emotion word under each classification and determine its Sentiment orientation;
Step S3, emotion tendency differentiates, specifically includes:The classification of the information resources of Sentiment orientation to be discriminated, according to this
The characteristic of information resources judges the type belonging to it;Chinese word segmentation processing, the emotion keyword in affection resources, from waiting to sentence
Emotion keyword is obtained in the information resources of other Sentiment orientation;Authoritative publisher judges, authority's hair is identified from the information resources
Cloth person, and the sentiment analysis result according to belonging to the polarity type of the authoritative publisher obtains the information resources;Social information feelings
Sense analysis, sentiment analysis is carried out to social category information;Universaling dictionary is analyzed, to the information or nothing that event category is general categories
The social category information that method obtains sentiment analysis result carries out Sentiment orientation analysis;Proprietary dictionary analysis, for thing in social information
Part classification carries out sentiment analysis for the information of proprietary classification.
Affection resources system is various dimensions, the resource hierarchy of more granularities, and wherein various dimensions are used to characterize in taxonomic hierarchies
Multiple class categories, more granularities are used to characterize the size of used characteristic particle size under certain dimension, for towards field
In each class categories, can the difference such as symbolization, word, phrase and sentence simultaneously according to the real needs of sentiment analysis
The character string of granularity and its corresponding Sentiment orientation in the category are as feature.
Selection for Sentiment orientation word, emotion keyword selected in same granularity level under different classes of dimension
It is incomplete same;Meanwhile for same emotion keyword, there can be different emotions to incline under different classification dimensions
To its specific Sentiment orientation should be selected according to the specific category dimension belonging to it.
The sentiment analysis method also includes building general affection resources and proprietary affection resources, wherein proprietary affection resources
Including authoritative account affection resources, social information viewpoint sentence affection resources, social information phrase affection resources, social information expression
Symbol affection resources and social information word affection resources.
Wherein, emotion tendency differentiate be according to the publishers of information resources whether be authoritative publisher, type whether be
Whether social information, content include comment, whether information is the multimedia messages such as video, whether property is news, affiliated event
Whether it is the attribute in totally seven dimensions such as the class categories of general categories and the information resources belonging in classification system, point
The other emotion tendency to information resources judges.
Wherein, it is the polarity that emotion is judged according to the authoritative account affection resources constructed that emotion tendency, which differentiates, for
The information of authoritative account issue is before calling social information sentiment analysis module or using general information sentiment analysis method
The Sentiment orientation of information can rapidly and accurately be judged.
Wherein, it is the information of proprietary classification point affiliated in classification system according to it for event category in social information
Class classification carries out sentiment analysis:Before the formally analysis process into patent dictionary analysis module, module can first judge the letter
The length of the core content text of breath, the Sentiment orientation that the information is thought if length is 0 is neutral, otherwise into the module
Analysis process;The viewpoint sentence in social information is extracted first, if viewpoint sentence be present, is united according to proprietary viewpoint sentence affection resources
The viewpoint sentence quantity hit in the information is counted, according to the feeling polarities of the viewpoint sentence set analysis of the hit information, is otherwise continued;
Secondly the phrase in extraction social information, the phrase quantity hit in the information, root are counted according to proprietary phrase affection resources
According to the feeling polarities of the phrase set analysis of the hit information, otherwise continue;Then the emoticon in social information, root are extracted
The emoticon quantity hit in the information is counted according to proprietary emoticon affection resources, according to the emoticon set of hit
The feeling polarities of the information are analyzed, are otherwise continued;The word in social information is finally extracted, according to proprietary word affection resources
The word quantity hit in the information is counted, the feeling polarities of the information are analyzed according to the set of words of hit, otherwise it is assumed that should
The feeling polarities of social information are unidentified polarity.
The present invention is described further with specific embodiment below with reference to accompanying drawings, with micro-blog information in embodiment
Illustrated for social information citing.
As shown in figure 1, a kind of various dimensions and more granularity sentiment analysis methods as one embodiment of the invention, including three
Individual part:Affection resources system construction, Sentiment orientation selected ci poem are selected, emotion tendency differentiates.Wherein, affection resources system construction walks
Suddenly, for the real needs according to sentiment analysis, and towards specific area text classification system, structure with the domain class
The affection resources that complicated variant system matches;Sentiment orientation selected ci poem selects step, for the real needs according to sentiment analysis, and structure
The classification system towards specific area text gone out, the affection resources to be matched with the domain class complicated variant system are built, for each
Class categories, according to the demand of reality can simultaneously the varigrained character string such as symbolization, word, phrase and sentence and its
The corresponding Sentiment orientation in the category is as feature;Emotion tendency discriminating step, using information to be analyzed as object,
Using the processing rule and its Sentiment orientation of process analysis shown in Fig. 2.
As shown in Fig. 2 the emotion tendency discriminating step of the embodiment of the present invention, according to the publisher of information resources whether be
Whether authoritative publisher and type are attribute in social information the two dimensions, start the emotion tendency to information resources
Judged, after parameter necessary to load dictionary resources necessary to sentiment analysis and determine, first determine whether information to be analyzed
Source, if authoritative account issue information, then be directly used as the information using the feeling polarities that the authoritative account releases news
Feeling polarities;Otherwise judge flow into information type, social information sentiment analysis is then called if social category information resource
Its emotion of module analysis, according to whether information resources content includes comment, whether information is the multimedia messages such as video, property is
It is no be news, affiliated event whether be the class categories etc. of general categories and the information resources belonging in classification system five
Attribute in dimension, continue to judge the emotion tendency of information resources, otherwise using general information sentiment analysis method
Analyze its emotion.
As shown in figure 3, social information sentiment analysis step, for carrying out sentiment analysis to social category information:First determine whether
With the presence or absence of information such as comments in social information, if comment information be present, the comment information is only handled as sentiment analysis
Object, the otherwise object using the full text of the information as sentiment analysis;Secondly according in social information whether be news, video,
Its emotion of the type analysis such as picture, otherwise continues;Finally whether the event category according to belonging to social information is general categories point
Different analysis modules is not called, if general categories, then the emotion pole of the information is calculated using universaling dictionary analysis module
Property, feeling polarities are otherwise calculated using proprietary dictionary analysis module.
As shown in figure 4, proprietary dictionary analytical procedure, for the information for event category in social information for proprietary classification
Class categories according to belonging to it in classification system carry out sentiment analysis:Before formally proprietary dictionary analysis process is entered,
The length of the core content text of the information can first be judged, the Sentiment orientation that the information is thought if length is 0 is neutrality, otherwise
Into the analysis process of the module.The viewpoint sentence in social information is extracted first, if viewpoint sentence be present, according to proprietary viewpoint
Sentence affection resources count the viewpoint sentence quantity hit in the information, according to the emotion pole of the viewpoint sentence set analysis of the hit information
Property, otherwise continue;Secondly the phrase in extraction social information, counts what is hit in the information according to proprietary phrase affection resources
Phrase quantity, according to the feeling polarities of the phrase set analysis of the hit information, otherwise continue;Then extract in social information
Emoticon, the emoticon quantity hit in the information is counted according to proprietary emoticon affection resources, according to hit
The feeling polarities of the emoticon set analysis information, otherwise continue;The word in social information is finally extracted, according to proprietary
Word affection resources count the word quantity hit in the information, and the emotion pole of the information is analyzed according to the set of words of hit
Property, otherwise it is assumed that the feeling polarities of the social information are unidentified polarity.
Particular embodiments described above, the purpose of the present invention, technical scheme and beneficial effect are carried out further in detail
Describe in detail bright, it should be understood that the foregoing is only the present invention specific embodiment, be not intended to limit the invention, it is all
Within the spirit and principles in the present invention, any modification, equivalent substitution and improvements done etc., the protection of the present invention should be included in
Within the scope of.
Claims (8)
1. a kind of various dimensions and more granularity sentiment analysis methods, wherein the publisher of the various dimensions characterization information resource whether be
Whether authoritative publisher, type are social information, whether content be video class multimedia messages, property comprising comment, information
Whether matter is news, whether affiliated event is that the class categories of general categories and the information resources belonging in classification system are total to
Attribute in seven dimensions;The size of used characteristic particle size, symbolization, word under more particle size Lambda characterization dimensions
Language, phrase and sentence totally four kinds of varigrained character strings and its corresponding Sentiment orientation in the classification;It is described
Sentiment analysis method comprises the following steps:
Step S1, affection resources are built, i.e., according to the real needs of sentiment analysis and the classification system of domain-oriented text, structure
Build the affection resources to be matched with the domain class complicated variant system;
Step S2, Sentiment orientation word is selected, i.e., according to the real needs of sentiment analysis, selected in institute's domain-oriented under each classification
Emotion word and determine its Sentiment orientation;
Step S3, differentiate emotion tendency, comprise the following steps:Judged according to the characteristic of the information resources of Sentiment orientation to be discriminated
Type belonging to it;Emotion keyword in the affection resources, from the information resources of the Sentiment orientation to be discriminated
Obtain emotion keyword;Authoritative publisher is identified from the information resources of the Sentiment orientation to be discriminated, and according to the authority
The polarity type of publisher obtains the sentiment analysis result belonging to described information resource;Sentiment analysis is carried out to social category information;
Event category for the information of general categories or can not be obtained by the social category information of sentiment analysis result carries out Sentiment orientation point
Analysis;For event category in social information sentiment analysis is carried out for the information of proprietary classification.
2. various dimensions according to claim 1 and more granularity sentiment analysis methods, wherein the affection resources be various dimensions,
The resource hierarchy of more granularities, wherein various dimensions are used to characterize multiple class categories in taxonomic hierarchies, and more granularities are used to characterize certain
The size of used characteristic particle size under dimension, for each class categories in institute's domain-oriented, according to sentiment analysis
Real needs can simultaneously symbolization, word, phrase and sentence totally four kinds of varigrained character strings and its it is corresponding
Sentiment orientation in the category is as feature.
3. various dimensions according to claim 1 and more granularity sentiment analysis methods, wherein the selection of the Sentiment orientation word
Standard is:Emotion keyword selected in same granularity level is incomplete same under different classes of dimension;Meanwhile for same
One emotion keyword, there are the Sentiment orientation corresponding with category dimension, its specific feelings under different classification dimensions
Sense tendency should be selected according to the specific category dimension belonging to it.
4. various dimensions according to claim 1 and more granularity sentiment analysis methods, wherein the sentiment analysis method is also wrapped
Include and construct general affection resources and proprietary affection resources.
5. various dimensions according to claim 4 and more granularity sentiment analysis methods, wherein the proprietary affection resources include
Authoritative account affection resources, social information viewpoint sentence affection resources, social information phrase affection resources, social information emoticon
Affection resources and social information word affection resources.
6. various dimensions according to claim 1 and more granularity sentiment analysis methods, wherein to social class described in step S3
Whether information was carried out in the step of sentiment analysis, be authoritative hair according to the publisher of the information resources of the Sentiment orientation to be discriminated
Cloth person, type whether be social information, content whether comprising comment, information whether be video class multimedia messages, property whether
Whether it is general categories and the information resources class categories affiliated in classification system totally seven dimensions for news, affiliated event
Attribute on degree, the emotion tendency of information resources is judged respectively.
7. various dimensions according to claim 1 and more granularity sentiment analysis methods, wherein to social class described in step S3
Information was carried out in the step of sentiment analysis, the polarity of emotion was judged according to the authoritative account affection resources constructed, for authority
The information of account issue can be quick before calling social information sentiment analysis or using general information sentiment analysis step
The Sentiment orientation of information is judged exactly.
8. various dimensions according to claim 1 and more granularity sentiment analysis methods, for social activity wherein described in step S3
The step of event category carries out sentiment analysis for the information of proprietary classification in information includes:
Before formally proprietary dictionary analysis process is entered, the length of the core content text of described information is judged, if length is 0
The Sentiment orientation for then thinking described information is neutrality, otherwise into the proprietary dictionary analysis process;
The viewpoint sentence in described information is extracted, if viewpoint sentence be present, the letter is counted according to proprietary viewpoint sentence affection resources
The viewpoint sentence quantity hit in breath, according to the feeling polarities of the viewpoint sentence set analysis described information of hit;
The phrase in described information is extracted, if phrase be present, is hit according in proprietary phrase affection resources statistics described information
Phrase quantity, according to the feeling polarities of the phrase set analysis described information of hit;
The emoticon in described information is extracted, if emoticon be present, institute is counted according to proprietary emoticon affection resources
The emoticon quantity hit in information is stated, according to the feeling polarities of the emoticon set analysis described information of hit;
The word in described information is extracted, if word be present, is hit according in proprietary word affection resources statistics described information
Word quantity, according to the set of words of hit analyze described information feeling polarities, otherwise it is assumed that the emotion pole of described information
Property is unidentified polarity.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510431992.8A CN105068988B (en) | 2015-07-21 | 2015-07-21 | Various dimensions and more granularity sentiment analysis methods |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510431992.8A CN105068988B (en) | 2015-07-21 | 2015-07-21 | Various dimensions and more granularity sentiment analysis methods |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105068988A CN105068988A (en) | 2015-11-18 |
CN105068988B true CN105068988B (en) | 2018-01-30 |
Family
ID=54498362
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510431992.8A Active CN105068988B (en) | 2015-07-21 | 2015-07-21 | Various dimensions and more granularity sentiment analysis methods |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105068988B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108121698A (en) * | 2017-11-29 | 2018-06-05 | 张宁 | Chinese text Emotion identification method and system |
CN109325112B (en) * | 2018-06-27 | 2019-08-20 | 北京大学 | A kind of across language sentiment analysis method and apparatus based on emoji |
CN113297934B (en) * | 2021-05-11 | 2024-03-29 | 国家计算机网络与信息安全管理中心 | Multi-mode video behavior analysis method for detecting Internet violence harmful scene |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2006125257A1 (en) * | 2005-05-25 | 2006-11-30 | Deborah Valerie Nicholas | A method of preventing or reducing the occurrence of psychosis and related symptoms using knowledge of actual mental processes in schizophrenia and other illness involving psychotic symptoms |
CN102236636A (en) * | 2010-04-26 | 2011-11-09 | 富士通株式会社 | Method and device for analyzing emotional tendency |
CN102486771A (en) * | 2009-11-30 | 2012-06-06 | 国际商业机器公司 | Method and system for releasing specified contents on webpage |
CN103365867A (en) * | 2012-03-29 | 2013-10-23 | 腾讯科技(深圳)有限公司 | Method and device for emotion analysis of user evaluation |
CN103559233A (en) * | 2012-10-29 | 2014-02-05 | 中国人民解放军国防科学技术大学 | Extraction method for network new words in microblogs and microblog emotion analysis method and system |
-
2015
- 2015-07-21 CN CN201510431992.8A patent/CN105068988B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2006125257A1 (en) * | 2005-05-25 | 2006-11-30 | Deborah Valerie Nicholas | A method of preventing or reducing the occurrence of psychosis and related symptoms using knowledge of actual mental processes in schizophrenia and other illness involving psychotic symptoms |
CN102486771A (en) * | 2009-11-30 | 2012-06-06 | 国际商业机器公司 | Method and system for releasing specified contents on webpage |
CN102236636A (en) * | 2010-04-26 | 2011-11-09 | 富士通株式会社 | Method and device for analyzing emotional tendency |
CN103365867A (en) * | 2012-03-29 | 2013-10-23 | 腾讯科技(深圳)有限公司 | Method and device for emotion analysis of user evaluation |
CN103559233A (en) * | 2012-10-29 | 2014-02-05 | 中国人民解放军国防科学技术大学 | Extraction method for network new words in microblogs and microblog emotion analysis method and system |
Also Published As
Publication number | Publication date |
---|---|
CN105068988A (en) | 2015-11-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
De Gibert et al. | Hate speech dataset from a white supremacy forum | |
CN106598944B (en) | A kind of civil aviaton's security public sentiment sentiment analysis method | |
Pons et al. | Impact of Corporate Social Responsibility in mining industries | |
CN109299271B (en) | Training sample generation method, text data method, public opinion event classification method and related equipment | |
WO2016197577A1 (en) | Method and apparatus for labelling comment information and computer device | |
CN103294778B (en) | A kind of method and system pushing information | |
CN109446404A (en) | A kind of the feeling polarities analysis method and device of network public-opinion | |
CN104268160A (en) | Evaluation object extraction method based on domain dictionary and semantic roles | |
CN104484336B (en) | A kind of Chinese comment and analysis method and its system | |
CN103077207B (en) | A kind of microblogging happy index analysis method and system | |
CN105447144B (en) | Microblogging forwarding visual analysis method and system based on big data analysis technology | |
CN107491435A (en) | Method and device based on Computer Automatic Recognition user feeling | |
CN102279890A (en) | Sentiment word extracting and collecting method based on micro blog | |
CN106126502A (en) | A kind of emotional semantic classification system and method based on support vector machine | |
CN108280057A (en) | A kind of microblogging rumour detection method based on BLSTM | |
CN105183717A (en) | OSN user emotion analysis method based on random forest and user relationship | |
CN113076735B (en) | Target information acquisition method, device and server | |
CN109634436A (en) | Association method, device, equipment and the readable storage medium storing program for executing of input method | |
CN104899335A (en) | Method for performing sentiment classification on network public sentiment of information | |
CN105068988B (en) | Various dimensions and more granularity sentiment analysis methods | |
CN108984514A (en) | Acquisition methods and device, storage medium, the processor of word | |
CN114547303A (en) | Text multi-feature classification method and device based on Bert-LSTM | |
CN118013045A (en) | Sentence emotion detection method and device based on artificial intelligence | |
CN107102984A (en) | A kind of Tibetan language microblog emotional sentiment classification method and system | |
CN110324278A (en) | Account main body consistency detecting method, device and equipment |
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 |