CN106951409A - A kind of network social intercourse media viewpoint tendency analysis system and method - Google Patents
A kind of network social intercourse media viewpoint tendency analysis system and method Download PDFInfo
- Publication number
- CN106951409A CN106951409A CN201710160543.3A CN201710160543A CN106951409A CN 106951409 A CN106951409 A CN 106951409A CN 201710160543 A CN201710160543 A CN 201710160543A CN 106951409 A CN106951409 A CN 106951409A
- Authority
- CN
- China
- Prior art keywords
- emotion
- viewpoint
- information
- word
- tendency
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000004458 analytical method Methods 0.000 title claims abstract description 72
- 238000000034 method Methods 0.000 title claims abstract description 32
- 230000008451 emotion Effects 0.000 claims abstract description 283
- 230000002452 interceptive effect Effects 0.000 claims abstract description 35
- 238000001914 filtration Methods 0.000 claims abstract description 24
- 241000239290 Araneae Species 0.000 claims abstract description 11
- 238000012216 screening Methods 0.000 claims description 54
- 238000000605 extraction Methods 0.000 claims description 30
- 230000002996 emotional effect Effects 0.000 claims description 16
- 230000007935 neutral effect Effects 0.000 claims description 15
- 238000012545 processing Methods 0.000 claims description 11
- 238000002372 labelling Methods 0.000 claims description 8
- 238000012163 sequencing technique Methods 0.000 claims description 8
- 210000003813 thumb Anatomy 0.000 claims description 8
- 108090000623 proteins and genes Proteins 0.000 claims description 7
- 238000012795 verification Methods 0.000 claims description 7
- 238000012549 training Methods 0.000 claims description 6
- 238000013527 convolutional neural network Methods 0.000 claims description 4
- 108010001267 Protein Subunits Proteins 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 3
- 238000012217 deletion Methods 0.000 claims description 3
- 230000037430 deletion Effects 0.000 claims description 3
- 238000010586 diagram Methods 0.000 claims description 3
- 230000010415 tropism Effects 0.000 claims description 3
- 238000010835 comparative analysis Methods 0.000 claims description 2
- 238000000151 deposition Methods 0.000 claims description 2
- 230000001537 neural effect Effects 0.000 claims description 2
- 235000013399 edible fruits Nutrition 0.000 claims 1
- 230000003993 interaction Effects 0.000 claims 1
- 238000004321 preservation Methods 0.000 claims 1
- 239000000047 product Substances 0.000 description 6
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 241000272183 Geococcyx californianus Species 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000003796 beauty Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 239000006227 byproduct Substances 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000009223 counseling Methods 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000013549 information retrieval technique Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 230000036651 mood Effects 0.000 description 1
- 238000003012 network analysis Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000020509 sex determination Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Business, Economics & Management (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a kind of network social intercourse media viewpoint tendency analysis system and method, including database and webpage capture module, viewpoint acquisition module, data filtering module, polarity check module；Database is used to store positive emotion cluster and negative sense emotion cluster；Webpage capture module is used to capture the comment Interactive Web Page of network social intercourse media by spiders；Viewpoint acquisition module is used for some viewpoint informations for reading user comment in comment Interactive Web Page；Data filtering module is used to screen viewpoint information；Polarity check module is used for the emotion tendency for analyzing the viewpoint information.The present invention can be for recognizing and extracting subjective information in the magnanimity initial data in the comment Interactive Web Page of network social intercourse media, and thus judge that the comment viewpoint of commentator holds attitude to something or other or certain event, prevent that substantial amounts of valuable viewpoint information is submerged, alleviate viewpoint information and cross the pressure brought, can quickly analyze the emotion tendency of viewpoint information.
Description
Technical field
The present invention relates to network information processing technical field, more particularly to a kind of network social intercourse media viewpoint sentiment classification
System and method.
Background technology
Internet has turned into one of most important platform of people's acquisition and exchange of information, internet and traditional media phase
Than, be one of the characteristics of maximum freely, it is open, in time, anyone can be by blog, forum, microblogging, space etc. in interconnection
Web realease viewpoint and comment.Network social intercourse media based on internet, are very easy to people on the internet to oneself
Policy, news, social event, focus personage and electric business product interested etc. deliver viewpoint and view.Network social intercourse media
The important carrier and channel with Information Communication are maintained as human society relation, national security and social development are generated
Far-reaching influence.
At present, exist in network social intercourse media and continue to produce the viewpoint comment information of magnanimity, they have important
Application value, has received the special attention of social each side.For example, government can understand people couple by public's comment information
Every policies and measures and the mood and attitude of accident, control public sentiment, so that correct decision-making is made in time, can also be to spy
Fixed high pressure crowd does sentiment analysis, so as to provide targetedly psychological counseling to them；Enterprise can be by product
Comment information, grasps the market response of product, so as to improve the performance of product in time or change sales tactics, lifts its society competing
Strive power；Masses by the evaluation informations of commodity, can understand the performance of product various aspects, with reference to the use body of other consumers
Test, so as to help to make their own purchase decision.But, the information comprising various viewpoints in network social intercourse media is just with exponential
Other speed increases, with containing much information, dispersiveness is strong, it is random it is strong, the features such as sentence structure is imperfect so that it is a large amount of valuable
The viewpoint information of value is submerged.At present, although the development of information retrieval technique alleviates this information overload to a certain extent
The pressure brought, but the most Bian keyword matchs of search engine, do not take into account viewpoint and sentiment analysis.Therefore, it is right
In a simply search, it is possible to return to a large amount of webpages comprising repetition and uncorrelated content, user needs to expend substantial amounts of
Time and efforts can be just found for something, someone, the viewpoint and emotion tendency information of something etc., this actual need with people
Ask also far apart, in addition, everyone is typically only capable to monitor the text message of certain several internet sites, information collects difficulty,
Easily ignore some hot informations.
In order to realize quick analysis and the statistics to internet public feelings information, existing patent publication No. is
CN104142913A discloses the method for discrimination and judgement system of word polarity, and the patent can realize the Sentiment orientation of word,
But analysis process is complex, the emotional orientation analysis degree of accuracy is not high；Meanwhile, existing patent publication No. is
CN101408883A discloses a kind of method for collecting network public feelings viewpoint, and it realizes viewpoint collection, and viewpoint emotionality judges very
Hardly possible adapts to the diversity of network social intercourse media, it is impossible to meet emotional orientation analysis of the user to internet public feelings information, therefore,
Be badly in need of exploitation one kind can express statistic internet public feelings information, and can interpolate that out the tendentious network society of viewpoint in information
Hand over media viewpoint sentiment classification method and system.
The content of the invention
In order to which the analysis process for solving existing media viewpoint analysis and collection technique presence is complex, emotion tendency point
The analysis degree of accuracy is not high, and viewpoint emotionality judges to be difficult the diversity for adapting to network social intercourse media, it is impossible to meet user to internet
The problems such as emotional orientation analysis of public feelings information, extract accurate the invention provides a kind of viewpoint, and can quickly analyze
The tendentious network social intercourse media viewpoint sentiment classification method and system of viewpoint.
Concrete technical scheme of the present invention is as follows：
The invention provides a kind of network social intercourse media viewpoint tendency analysis system, including database and with the data
Webpage capture module that storehouse communicates, viewpoint acquisition module, data filtering module, polarity check module；The database is used for
The positive emotion cluster of storage and negative sense emotion cluster, if being stored with the positive emotion cluster and the negative sense emotion cluster
Dry emotionality word or word, emotionality word or word include adjective, verb, noun, adverbial word；
The webpage capture module is used to capture the comment Interactive Web Page of network social intercourse media by spiders,
The network social intercourse media include microblogging, wechat, blog, forum, blog, transaction platform；
The viewpoint acquisition module is used for some viewpoint informations for reading user comment in the comment Interactive Web Page；
The data filtering module is used to screen the viewpoint information, and will be with being commented in the network social intercourse media
Deleted by the unrelated viewpoint information of theme；
The polarity check module is used to extract some Sentiment orientation sex factors in the viewpoint information, and the emotion is inclined
It is emotionality word or word to sex factor, and by the Sentiment orientation sex factor and the positive emotion cluster and the negative sense feelings
The emotionality word or word felt in cluster carry out matching degree judgement, count some emotion tendencies in the viewpoint information
The factor is belonging respectively to the quantity of the positive emotion cluster and the negative sense emotion cluster, and the sight is gone out by volume comparison analysis
The emotion tendency of point information, while assigning polarity label to the viewpoint information, the polarity label includes positive emotion letter
Breath and negative sense emotion information.
Further, the data filtering module includes subject gene extraction unit, the correlation model construction list communicated
Member, keyword extraction unit, degree of correlation judging unit, filtering examination ＆ verification unit,
If the subject gene extraction unit is used to extract related to the comment theme in the comment Interactive Web Page
Dry key factor, the key factor includes the keyword commented in theme, for commenting on describing for the comment theme
Word；
The correlation model, which builds unit, to be used for some key factors as being training sample to convolutional Neural net
Network model is trained, and sets up relevance model；
The keyword extraction unit is used to extract the keyword in the viewpoint information；
The degree of correlation judging unit is used to input the keyword into the relevance model to be trained, and
Go out the similarity output valve of the viewpoint information and the comment theme；
The filtering examination ＆ verification unit is used for the viewpoint that the similarity output valve is less than to default similarity threshold values
Information deletion, and the viewpoint information that the similarity output valve is more than or equal to the similarity threshold values is sent to described
Polarity check module.
Further, the analysis system also includes communicating with the data filtering module and the polarity check module
The viewpoint information screening module of news, the screening conditions that the viewpoint information screening module includes communicating preset unit, screening list
Member, sequencing unit, the screening conditions, which preset unit, is used for the default screening factor, and the screening factor includes period, follow-up
At least one in number, thumb up number, the screening unit is used in the comment Interactive Web Page according to the screening factor pair
Some viewpoint informations are screened, and the viewpoint information filtered out is sent to the sequencing unit, the sequence
Unit be used to being ranked up the viewpoint information filtered out by high order on earth according to the screening factor and send to
The polarity check module.
Further, the polarity check module includes the threshold preset unit, word extraction unit, matching degree communicated
Judging unit, emotion tendency processing unit, emotional orientation analysis unit, label for labelling unit；
The threshold preset unit be used for tendency threshold value default to the emotionality word or word in the positive emotion cluster+
F, while for being inclined to threshold value-F to the emotionality word or word in the negative sense emotion cluster are default, wherein, F is integer；
The word extraction unit is used to extract some Sentiment orientation sex factors in the viewpoint information；
The matching degree judging unit includes judgment sub-unit, retrieval subelement, separately deposits subelement, the judgment sub-unit
For by the emotionality word in the Sentiment orientation sex factor of extraction and the positive emotion cluster and the negative sense emotion cluster or
Word carries out matching degree judgement, and positive emotion word or word are belonged to when Sentiment orientation sex factor belongs to the positive emotion cluster
Language, belongs to negative sense emotion word or word when Sentiment orientation sex factor belongs to the negative sense emotion cluster, works as emotion tendency
When the factor had both been not belonging to the positive emotion cluster or had been not belonging to the negative sense emotion cluster, then by the Sentiment orientation sex factor
Send to the retrieval subelement, the retrieval subelement is by retrieving the Sentiment orientation sex factor in the network social intercourse matchmaker
Sentiment orientation information in the historical events of body, and send to it is described it is another deposit subelement, it is described it is another deposit subelement be used for will described in
Sentiment orientation sex factor is preserved to the positive emotion cluster or the negative sense emotion cluster according to its Sentiment orientation information correspondence
In；
Some emotion tendency factors difference that the emotion tendency processing unit is used to count in the viewpoint information
Belong to the quantity of the positive emotion cluster and the negative sense emotion cluster, and the viewpoint information is calculated by below equation
Emotion tendency value W：
W=N × (+F)+M × (- F)；
Wherein, W is the emotion tendency value W of the viewpoint information；N belongs to the forward direction for the Sentiment orientation sex factor
The quantity of emotion cluster；M is the quantity that the Sentiment orientation sex factor belongs to the negative sense emotion cluster, and F is tendency threshold value；
The emotional orientation analysis unit is used for the feelings that the viewpoint information is analyzed according to the emotion tendency value W
Feel tendentiousness, be positive emotion when the emotion tendency value W is more than or equal to tendency threshold value+F；When the Sentiment orientation
Property value W when being less than or equal to tendency threshold value-F, as negative sense emotion；When the emotion tendency value W is equal to 0, disposition in being
Sense；
The viewpoint information that the label for labelling unit is used to the emotional orientation analysis element analysis is assigned
Polarity label, the polarity label includes positive emotion information, negative sense emotion information and neutral emotion information, and sends to described
Viewpoint polarity statistical module.
It is preferred that, the analysis system also include the viewpoint polarity statistical module that is communicated with the polarity check module and
Result display module, the viewpoint polarity statistical module is used for according to some viewpoint informations in the comment Interactive Web Page
Polarity label counts the quantity of the positive emotion information, the negative sense emotion information and the neutral emotion information respectively,
And send to the result display module；The result display module is used for the positive emotion information, the negative sense emotion
Information and the quantity of the neutral emotion information are drawn viewpoint analysis block diagram and sent to the database and preserves.
It is preferred that, the webpage capture module includes network address acquiring unit, the webpage capture unit, the network address communicated
Acquiring unit is used for the URL network address for obtaining the network social intercourse media, and the webpage capture unit is used to grab using spiders
Device is taken to capture the comment Interactive Web Page of the network social intercourse media.
Present invention also offers a kind of network social intercourse media viewpoint sentiment classification method, the analysis method includes following
Step：
S1, by spiders the comment Interactive Web Page of network social intercourse media is captured, the network social intercourse media
Including microblogging, wechat, blog, forum, blog, transaction platform；
S2, some viewpoint informations for reading user comment in the comment Interactive Web Page；
S3, the viewpoint information screened, and will it is unrelated with commenting on theme in the network social intercourse media described in
Viewpoint information is deleted；
S4, some Sentiment orientation sex factors extracted in the viewpoint information, the Sentiment orientation sex factor include emotion
Tendentiousness word or emotion tendency word, and by the Sentiment orientation sex factor and the positive emotion cluster and the negative sense emotion
Emotionality word in cluster or word carry out matching degree judgement, at the same count some emotion tendencies in the viewpoint information because
Son is belonging respectively to the quantity of the positive emotion cluster and the negative sense emotion cluster, and the viewpoint is gone out by volume comparison analysis
The emotion tendency of information, while assigning polarity label to the viewpoint information, the polarity label includes positive emotion information
With negative sense emotion information.
Further, in step S3, the viewpoint information is screened, and will be with being commented in the network social intercourse media
The unrelated viewpoint information of theme is deleted, and specific method is：
Some key factors related to the comment theme in S3-1, the extraction comment Interactive Web Page；
S3-2, some key factors are trained as training sample to convolutional neural networks model, set up
Relevance model；
S3-3, the keyword extracted in the viewpoint information；
S3-4, the keyword inputted into the relevance model be trained, and draw the viewpoint information with
The similarity output valve of the comment theme；
S3-5, the viewpoint information by the similarity output valve less than default similarity threshold values are deleted.
Further, the step S4 specifically includes following methods：
S4-1, tendency threshold value+F default to the emotionality word or word in the positive emotion cluster, while for institute
The default tendency threshold value-F of emotionality word or word in negative sense emotion cluster is stated, wherein, F is integer；
S4-2, some Sentiment orientation sex factors extracted in the viewpoint information；；
S4-3, by the feelings in the Sentiment orientation sex factor of extraction and the positive emotion cluster and the negative sense emotion cluster
Perceptual word or word carry out matching degree judgement, belong to positive feelings when Sentiment orientation sex factor belongs to the positive emotion cluster
Feel word or word, belong to negative sense emotion word or word when Sentiment orientation sex factor belongs to the negative sense emotion cluster；
Some emotion tendency factors in S4-4, the statistics viewpoint information be belonging respectively to the positive emotion cluster and
The quantity of the negative sense emotion cluster, and calculate by below equation the emotion tendency value W of the viewpoint information：
W=N × (+F)+M × (- F)；
Wherein, W is the emotion tendency value W of the viewpoint information；N belongs to the forward direction for the Sentiment orientation sex factor
The quantity of emotion cluster；M is the quantity that the Sentiment orientation sex factor belongs to the negative sense emotion cluster, and F is tendency threshold value；
S4-5, the emotion tendency for analyzing according to the emotion tendency value W viewpoint information, when the emotion is inclined
It is positive emotion when tropism value W is more than or equal to tendency threshold value+F；When the emotion tendency value W be less than or equal to tendency threshold value-
During F, as negative sense emotion；It is neutral emotion when the emotion tendency value W is equal to 0；
S4-6, assign polarity label to the viewpoint information that analyzes, the polarity label include positive emotion information,
Negative sense emotion information and neutral emotion information.
It is preferred that, step S3 also includes screening the viewpoint information, and screening technique is：
1. the screening factor is preset, the screening factor includes at least one in period, follow-up number, thumb up number；
2. some viewpoint informations in the comment Interactive Web Page according to the screening factor pair are screened；
3. the viewpoint information filtered out is ranked up according to the screening factor by high order on earth.
Beneficial effects of the present invention are as follows：The system and method that the present invention is provided can be directed to the comment of network social intercourse media
Recognized in magnanimity initial data in Interactive Web Page and extract subjective information, and thus judge the comment viewpoint of commentator to something
Thing or certain event hold attitude, effectively prevent that substantial amounts of valuable viewpoint information is submerged, alleviate viewpoint information and cross carrier band
The pressure come, can quickly analyze the emotion tendency of viewpoint information, effectively be easy to network social intercourse media to the pipe of viewpoint information
Reason and classification, with high accuracy, emotionality analyze speed is very fast, practical.
Brief description of the drawings
Fig. 1 is a kind of structured flowchart of network social intercourse media viewpoint tendency analysis system described in embodiment 1；
Fig. 2 is data filtering module in a kind of network social intercourse media viewpoint tendency analysis system described in embodiment 2
Structured flowchart；
Fig. 3 is viewpoint information screening mould in a kind of network social intercourse media viewpoint tendency analysis system described in embodiment 2
The structured flowchart of block；
Fig. 4 is polarity check module in a kind of network social intercourse media viewpoint tendency analysis system described in embodiment 3
Structured flowchart；
Fig. 5 is a kind of structured flowchart of network social intercourse media viewpoint tendency analysis system described in embodiment 3；
Fig. 6 is webpage capture module in a kind of network social intercourse media viewpoint tendency analysis system described in embodiment 3
Structured flowchart；
Fig. 7 is a kind of flow chart of network social intercourse media viewpoint sentiment classification method described in embodiment 4；
Fig. 8 be embodiment 5 described in a kind of network social intercourse media viewpoint sentiment classification method in step 3 operating process
Figure.
Wherein：1st, database；101st, positive emotion cluster；102nd, negative sense emotion cluster；2nd, webpage capture module；3rd, viewpoint
Acquisition module；4th, data filtering module；401st, subject gene extraction unit；402nd, correlation model builds unit；403rd, keyword
Extraction unit；404th, degree of correlation judging unit；405th, filtering examination ＆ verification unit；5th, polarity check module；501st, threshold preset unit；
502nd, word extraction unit；503rd, matching degree judging unit；504th, emotion tendency processing unit；505th, emotional orientation analysis
Unit；506th, label for labelling unit；6th, viewpoint information screening module；601st, screening conditions preset unit；602nd, screening unit；
603rd, sequencing unit；7th, viewpoint polarity statistical module；8th, result display module.
Embodiment
The present invention is described in further detail with following examples below in conjunction with the accompanying drawings.
Embodiment 1
As shown in figure 1, the embodiment of the present invention 1 provides a kind of network social intercourse media viewpoint tendency analysis system, this is
It is the product of multiple subject convergences to unite to opining mining and emotional orientation analysis, and it is related to artificial intelligence, linguistics, engineering
The multiple fields such as habit, data mining, information retrieval, mainly recognize from magnanimity initial data and extract subjective information, and thus
Judge that commentator holds attitude to something or other or certain event.The system includes database 1 and the net communicated with the database 1
Page handling module 2, viewpoint acquisition module 3, data filtering module 4, polarity check module 5, database 1 can be corpus, institute
Stating database 1 is used to store positive emotion cluster 101 and negative sense emotion cluster 102, the positive emotion cluster 101 and described negative
Be stored with some emotionality words or word into emotion cluster 102, emotionality word or word include adjective, verb, noun,
Word or word in adverbial word or escape word, database 1 can constantly update, emotionality word in positive emotion cluster 101 or
Word such as it is good, outstanding, support, praise, like, rod, beauty, emotionality word or word in negative sense emotion cluster 102 are for example
Disagreeable, dislike, or not it is no, painful, poor, ugly etc., commendation adjective can preferably, it is beautiful, beautiful etc.；Derogatory sense adjective can be with
To be poor, ugly, ugly, ugly etc.；Adverbial word can be greatly, very more to be fabulous；Commendation verb can be promotion, promotion, point
Praise etc.；Derogatory sense verb can be destruction, weak, disappointing etc.；Commendation noun can for happy, advantage, like, it is disagreeable etc.；
Derogatory sense noun can be scumbag, shortcoming etc.；Escape word can be do not have, it is not no, not etc..Corpus is used for the word for providing standard
Storehouse, the emotionality of viewpoint information is can interpolate that by the corpus.
The present invention quickly analyzes the viewpoint information of commentator by system, solve it is artificial need to take a substantial amount of time and
The problem of energy judges information emotion tendency, is effectively automatically analyzed by system, concluded and reasoning, is realized quick to viewpoint
The analysis and classification of the emotion tendency of information, improve analysis and conclude efficiency, practical.
The webpage capture module 2 is used to grab the comment Interactive Web Page of network social intercourse media by spiders
Take, the network social intercourse media include microblogging, wechat, blog, forum, blog, transaction platform；(be otherwise known as web crawlers net
Page spider, network robot is more frequent to be referred to as webpage follower in the middle of FOAF communities), it is a kind of according to certain rule
Then, the program or script of web message are automatically captured, the rarely needed name of other also has ant, from dynamic search
Draw, simulation program or worm.
The viewpoint acquisition module 3 is used for some viewpoint informations for reading user comment in the comment Interactive Web Page；See
Point information gathering is extracted using RoadRunner algorithms to the user comment information webpage of crawl.
For analysis time of the system of saving to comment information, the data filtering module 4 is used for the viewpoint information
Screened, and the viewpoint information unrelated with commenting on theme in the network social intercourse media is deleted；Will be independent of comment
The viewpoint of theme is deleted first, effectively mitigates system pressure.
Polarity check is carried out to the viewpoint information after screening, the polarity check module 5 is used to extract the viewpoint information
Interior some Sentiment orientation sex factors, the Sentiment orientation sex factor is emotionality word or word, and by the emotion tendency
The factor carries out matching degree with the emotionality word in the positive emotion cluster 101 and the negative sense emotion cluster 102 or word and sentenced
Disconnected, some emotion tendency factors counted in the viewpoint information are belonging respectively to the positive emotion cluster 101 and institute
The quantity of negative sense emotion cluster 102 is stated, the emotion tendency of the viewpoint information is gone out by volume comparison analysis, while to described
Viewpoint information assigns polarity label, and the polarity label includes positive emotion information and negative sense emotion information.
Specific analytical method is to judge Sentiment orientation one by one to the Sentiment orientation sex factor in a viewpoint information first
Property, it is more to judge which kind of emotion tendency word then according to the quantity of different emotions Tendency Factor in viewpoint information, so that
Determine to belong in the emotion tendency of the viewpoint information, such as viewpoint information emotion tendency of positive emotion cluster 101 because
Son is more than the Sentiment orientation sex factor for belonging to negative sense emotion cluster 102, then the viewpoint information belongs to positive emotion information, therefore,
The label of positive emotion information is assigned, is easy to observation and later stage statistics.
Embodiment 2
The embodiment of the present invention 2 further defines the structure of analysis system on the basis of embodiment 1, effectively increases and is
The analysis efficiency united to viewpoint information.
As shown in Fig. 2 in order to reduce the workload of system, it is necessary to be filtered to the viewpoint information of collection, to prevent unrelated
The viewpoint information of comment theme carries out emotional orientation analysis again, and explanation, the data filtering are needed further exist for for this
Module 4 includes the subject gene extraction unit 401 communicated, correlation model construction unit 402, keyword extraction unit 403, phase
Pass degree judging unit 404, filtering examination ＆ verification unit 405.
The subject gene extraction unit 401 is used to extract related to the comment theme in the comment Interactive Web Page
Some key factors, the key factor includes the keyword in the comment theme, the shape for commenting on the comment theme
Hold word；Extract comment theme in key factor when, extract can it is representative and can accurate description comment theme
Viewpoint information, if occurring and the unrelated information of comment theme, directly deletion.
The correlation model, which builds unit 402, to be used for some key factors as being training sample to convolutional Neural
Network model is trained, and sets up relevance model；The keyword extraction unit 403 is used to extract in the viewpoint information
Keyword；The degree of correlation judging unit 404 is used to input the keyword into the relevance model to be trained, and
Draw the similarity output valve of the viewpoint information and the comment theme；The filtering examination ＆ verification unit 405 is used for the phase
The viewpoint information like degree output valve less than default similarity threshold values is deleted, and the similarity output valve is more than or waited
Sent in the viewpoint information of the similarity threshold values to the polarity check module 5.Can be quick by relevance model
Whether related to comment theme analyze viewpoint information, retain if related, directly deleted if uncorrelated, improve system to having
The analysis efficiency of viewpoint information is imitated, it is practical.
As shown in figure 3, being carried out in order to the higher viewpoint information of focus degree or more representational viewpoint information
Further defined in analysis, the technical program, the analysis system also includes and the data filtering module 4 and the pole
Property the viewpoint information screening module 6 that communicates of analysis module 5, the viewpoint information screening module 6 includes the screening communicated
Condition presets unit 601, screening unit 602, sequencing unit 603, the screening conditions preset unit 601 be used for default screening because
Son, the screening factor includes at least one in period, follow-up number, thumb up number, and the screening unit 602 is used for according to institute
Some viewpoint informations stated in comment Interactive Web Page described in screening factor pair are screened, and by the viewpoint filtered out
Information is sent to the sequencing unit 603, and the sequencing unit 603 is used for the viewpoint information that will be filtered out according to the sieve
The factor is selected to be ranked up and sent to the polarity check module 5 by high order on earth.
The passage time setting of section can select the viewpoint information of some period, can be by the viewpoint beyond the period
Information is rejected, while according to the screening of follow-up number or thumb up number, the higher viewpoint information of temperature can be picked out, to the category information
Analysis with more emotional orientation analysis representativeness, it is practical, effectively shorten the time of network analysis, can pass through
The system quickly understands the tendentiousness of commentator.
Embodiment 3
The embodiment of the present invention 3 further defines how system carries out polarity check on the basis of embodiment 1, improves
The convenience of viewpoint information polarity check.
As shown in Figure 4, it is necessary to further limit, the polarity check module 5 includes the threshold preset list communicated
Member 501, word extraction unit 502, matching degree judging unit 503, emotion tendency processing unit 504, emotional orientation analysis
Unit 505, label for labelling unit 506；
The threshold preset unit 501, which is used to preset the emotionality word or word in the positive emotion cluster 101, to incline
To threshold value+F, while for being inclined to threshold value-F to the emotionality word or word in the negative sense emotion cluster 102 are default, wherein, F
For integer.Setting by being inclined to threshold value can be used in the emotion tendency value of computed view point information, consequently facilitating judging viewpoint
The emotion tendency of information.
The word extraction unit 502 is used to extract some Sentiment orientation sex factors in the viewpoint information；Word is carried
During taking, it is initially used for handling viewpoint information, participle and part-of-speech tagging processing is carried out first, punctuate symbol is then carried out
Number processing, expression meet processing and stop words processing, finally propose viewpoint information in adjective, verb, noun, adverbial word or turn
Adopted word is used as Sentiment orientation sex factor.
The matching degree judging unit 503 includes judgment sub-unit, retrieval subelement, separately deposits subelement, judgement
Unit is used in the Sentiment orientation sex factor of extraction and the positive emotion cluster 101 and the negative sense emotion cluster 102
Emotionality word or word carry out matching degree judgement, belong to when Sentiment orientation sex factor belongs to the positive emotion cluster 101
Positive emotion word or word, belong to when Sentiment orientation sex factor belongs to the negative sense emotion cluster 102 negative sense emotion word or
Word, the negative sense emotion cluster 102 is also not belonging to when Sentiment orientation sex factor neither belongs to the positive emotion cluster 101
When, then the Sentiment orientation sex factor is sent to the retrieval subelement, the retrieval subelement is by retrieving the emotion
Sentiment orientation information of the Tendency Factor in the historical events of the network social intercourse media, and transmission separately deposits sub single to described
Member, the another subelement of depositing is used to according to its Sentiment orientation information correspondence preserve the Sentiment orientation sex factor to the forward direction
In emotion cluster 101 or the negative sense emotion cluster 102；When the Sentiment orientation sex factor extracted in viewpoint information is not admitted to
When positive the emotion cluster 101 or the negative sense emotion cluster 102 of database 1, it can be gone through according to the retrieval of network big data
Occur the word in historical event part, to judge its polarity, and be stored in database 1, realize the renewal of database 1, it is convenient with
Polarity judging afterwards.
The emotion tendency processing unit 504 is used to count some emotion tendency Factor minutes in the viewpoint information
Do not belong to the quantity of the positive emotion cluster 101 and the negative sense emotion cluster 102, and the sight is calculated by below equation
The emotion tendency value W of point information：
W=N × (+F)+M × (- F)；
Wherein, W is the emotion tendency value W of the viewpoint information；N belongs to the forward direction for the Sentiment orientation sex factor
The quantity of emotion cluster 101；M is the quantity that the Sentiment orientation sex factor belongs to the negative sense emotion cluster 102, and F is tendency
Threshold value；
The emotional orientation analysis unit 505 is used to analyze the viewpoint information according to the emotion tendency value W
Emotion tendency, be positive emotion when the emotion tendency value W is more than or equal to tendency threshold value+F；When the emotion
When tendentiousness value W is less than or equal to tendency threshold value-F, as negative sense emotion；When the emotion tendency value W is equal to 0, in being
Disposition sense.
By the calculating of above-mentioned emotion tendency value, the polarity of viewpoint information can effectively be judged, such as viewpoint
Information is：As smart as a new pin, quality-high and inexpensive, Sentiment orientation sex factor is beautiful, inexpensive including beautiful, thing, passes through matching degree judging unit
After 503 pairs of these three Sentiment orientation sex factors judge, 3 belong to positive emotion cluster 101, then this viewpoint information
Emotionality value=3 × (+F)+0 × (- F)=+ 3F, because+3F is more than tendency threshold value+F, then the viewpoint information belongs to positive feelings
Sense.
The label for labelling unit 506 is used for the viewpoint letter analyzed to the emotional orientation analysis unit 505
Breath assigns polarity label, and the polarity label includes positive emotion information, negative sense emotion information and neutral emotion information, and sends
To the viewpoint polarity statistical module 7.The viewpoint information after analysis can be labeled by label for labelling unit 506, just
In statistics, the polarity for getting information about viewpoint information can be compared.
As shown in figure 5, being counted for convenience to a large amount of comment informations, preferred in the technical program to define, institute
Viewpoint polarity statistical module 7 and result display module 8 that analysis system also includes communicating with the polarity check module 5 are stated,
The viewpoint polarity statistical module 7 is used to be distinguished according to the polarity label of some viewpoint informations in the comment Interactive Web Page
The quantity of the positive emotion information, the negative sense emotion information and the neutral emotion information is counted, and is sent to described
Result display module 8；The result display module 8 is used for the positive emotion information, the negative sense emotion information and described
The quantity of neutral emotion information is drawn viewpoint analysis block diagram and sent to the database 1 and preserves.Pass through result display module 8
The emotion tendency of a large amount of comment informations can be intuitively observed, the statistics for the comment viewpoint being effectively easy to and management.
As shown in Figure 6, it is preferred that the webpage capture module 2 includes network address acquiring unit, the webpage capture list communicated
Member, the network address acquiring unit is used for the URL network address for obtaining the network social intercourse media, and the webpage capture unit is used to utilize
Spiders grabber captures the comment Interactive Web Page of the network social intercourse media.Comment can effectively be obtained by crawler technology
Interactive Web Page, consequently facilitating the collection of viewpoint information data.
Embodiment 4
As shown in fig. 7, the embodiment of the present invention 4 provides a kind of network social intercourse media viewpoint sentiment classification method, it is described
Analysis method comprises the following steps：
S1, by spiders the comment Interactive Web Page of network social intercourse media is captured, the network social intercourse media
Including microblogging, wechat, blog, forum, blog, transaction platform；User comment is some in S2, the reading comment Interactive Web Page
Viewpoint information；S3, the viewpoint information screened, and will it is unrelated with commenting on theme in the network social intercourse media described in
Viewpoint information is deleted；S4, some Sentiment orientation sex factors extracted in the viewpoint information, the Sentiment orientation sex factor include
Emotion tendency word or emotion tendency word, and by the Sentiment orientation sex factor and the positive emotion cluster 101 and described
Emotionality word or word in negative sense emotion cluster 102 carry out matching degree judgement, while counting some in the viewpoint information
The emotion tendency factor is belonging respectively to the quantity of the positive emotion cluster 101 and the negative sense emotion cluster 102, passes through quantity
Comparative analysis goes out the emotion tendency of the viewpoint information, while assigning polarity label, the polarity mark to the viewpoint information
Label include positive emotion information and negative sense emotion information.
The analysis method that the embodiment of the present invention 4 is provided can not only effectively crawl the viewpoint information in comment Interactive Web Page,
And the Sentiment orientation sex factor in information can be extracted, and polarity judgement is carried out to Sentiment orientation sex factor, so that
Realize and the emotion tendency of viewpoint information is judged.
Embodiment 5
The embodiment of the present invention 5 is further defined on the basis of embodiment 4 to method.
As shown in Figure 8, it is necessary to limit, when carrying out Sentiment orientation sex determination for certain complete viewpoint information, lead to
Often it is divided into the information related to theme and the information unrelated with theme, in order to analyze effective information, in the technical program
Define, in step S3, the viewpoint information is screened, and will be unrelated with commenting on theme in the network social intercourse media
The viewpoint information is deleted, and specific method is：
Some key factors related to the comment theme in S3-1, the extraction comment Interactive Web Page；
S3-2, some key factors are trained as training sample to convolutional neural networks model, set up
Relevance model；
S3-3, the keyword extracted in the viewpoint information；
S3-4, the keyword inputted into the relevance model be trained, and draw the viewpoint information with
The similarity output valve of the comment theme；
S3-5, the viewpoint information by the similarity output valve less than default similarity threshold values are deleted.
Deleted by the viewpoint information to unrelated subject matter, the screening to a large amount of viewpoint informations improved in the technical program,
Improve analysis efficiency.
Further, further specifically define that the step S4 specifically includes following methods in the technical program：
S4-1, tendency threshold value+F default to the emotionality word or word in the positive emotion cluster 101, at the same for pair
The default tendency threshold value-F of emotionality word or word in the negative sense emotion cluster 102, wherein, F is integer；
S4-2, some Sentiment orientation sex factors extracted in the viewpoint information；；
S4-3, by the Sentiment orientation sex factor of extraction and the positive emotion cluster 101 and the negative sense emotion cluster 102
Interior emotionality word or word carry out matching degree judgement, when Sentiment orientation sex factor belongs to the positive emotion cluster 101 i.e.
Belong to positive emotion word or word, belong to negative sense emotion when Sentiment orientation sex factor belongs to the negative sense emotion cluster 102
Word or word；
Some emotion tendency factors in S4-4, the statistics viewpoint information are belonging respectively to the positive emotion cluster
101 and the quantity of the negative sense emotion cluster 102, and calculate by below equation the emotion tendency value W of the viewpoint information：
W=N × (+F)+M × (- F)；
Wherein, W is the emotion tendency value W of the viewpoint information；N belongs to the forward direction for the Sentiment orientation sex factor
The quantity of emotion cluster 101；M is the quantity that the Sentiment orientation sex factor belongs to the negative sense emotion cluster 102, and F is tendency
Threshold value；
S4-5, the emotion tendency for analyzing according to the emotion tendency value W viewpoint information, when the emotion is inclined
It is positive emotion when tropism value W is more than or equal to tendency threshold value+F；When the emotion tendency value W be less than or equal to tendency threshold value-
During F, as negative sense emotion；It is neutral emotion when the emotion tendency value W is equal to 0；
S4-6, assign polarity label to the viewpoint information that analyzes, the polarity label include positive emotion information,
Negative sense emotion information and neutral emotion information.
It is preferred that, it is preferred in order to improve in the analysis efficiency to viewpoint information, the technical program, first to some viewpoints
Information is screened and filtered, and step S3 also includes screening the viewpoint information, and screening technique is：
1. the screening factor is preset, the screening factor includes at least one in period, follow-up number, thumb up number；2. root
Screened according to some viewpoint informations in comment Interactive Web Page described in the screening factor pair；3. described in filtering out
Viewpoint information is ranked up according to the screening factor by high order on earth.
Representative viewpoint information is analyzed in order to improve, the method that the technical program is provided being capable of pin
Different time sections and the higher viewpoint information of comment temperature are screened, according to comment time or comment temperature after screening
(thumb up number, follow-up number) is ranked up, so that the convenient information higher for comment temperature is analyzed, has more information analysis
It is representative.
The present invention is not limited to above-mentioned preferred forms, and anyone can show that other are various under the enlightenment of the present invention
The product of form, however, make any change in its shape or structure, it is every that there is skill identical or similar to the present application
Art scheme, is within the scope of the present invention.
Claims (10)
1. a kind of network social intercourse media viewpoint tendency analysis system, it is characterised in that including database (1) and with the data
Webpage capture module (2) that storehouse (1) communicates, viewpoint acquisition module (3), data filtering module (4), polarity check module (5)；
The database (1) is used to store positive emotion cluster (101) and negative sense emotion cluster (102), the positive emotion cluster
(101) and in the negative sense emotion cluster (102) be stored with some emotionality words or word, and emotionality word or word include shape
Hold word, verb, noun, adverbial word；
The webpage capture module (2) is used to capture the comment Interactive Web Page of network social intercourse media by spiders,
The network social intercourse media include microblogging, wechat, blog, forum, blog, transaction platform；
The viewpoint acquisition module (3) is used for some viewpoint informations for reading user comment in the comment Interactive Web Page；
The data filtering module (4) is used to screen the viewpoint information, and will be with being commented in the network social intercourse media
Deleted by the unrelated viewpoint information of theme；
The polarity check module (5) is used to extract some Sentiment orientation sex factors in the viewpoint information, and the emotion is inclined
It is emotionality word or word to sex factor, and by the Sentiment orientation sex factor and the positive emotion cluster (101) and described
Emotionality word or word in negative sense emotion cluster (102) carry out matching degree judgement, count some institutes in the viewpoint information
The quantity that the emotion tendency factor is belonging respectively to the positive emotion cluster (101) and the negative sense emotion cluster (102) is stated, is led to
The emotion tendency that volume comparison analysis goes out the viewpoint information is crossed, while polarity label is assigned to the viewpoint information, it is described
Polarity label includes positive emotion information and negative sense emotion information.
2. network social intercourse media viewpoint tendency analysis system as claimed in claim 1, it is characterised in that the data filtering
Module (4) includes the subject gene extraction unit (401), correlation model construction unit (402), keyword extraction unit communicated
(403), degree of correlation judging unit (404), filtering examination ＆ verification unit (405),
The subject gene extraction unit (401) is if related to the comment theme in the comment Interactive Web Page for extracting
Dry key factor, the key factor includes the keyword commented in theme, for commenting on describing for the comment theme
Word；
The correlation model, which builds unit (402), to be used for some key factors as being training sample to convolutional Neural net
Network model is trained, and sets up relevance model；
The keyword extraction unit (403) is used to extract the keyword in the viewpoint information；
The degree of correlation judging unit (404) is used to input the keyword into the relevance model to be trained, and
Draw the similarity output valve of the viewpoint information and the comment theme；The filtering examination ＆ verification unit (405) is used for will be described
Similarity output valve is deleted less than the viewpoint information of default similarity threshold values, and the similarity output valve is more than or
The viewpoint information equal to the similarity threshold values is sent to the polarity check module (5).
3. network social intercourse media viewpoint tendency analysis system as claimed in claim 1, it is characterised in that the analysis system
Also include the viewpoint information screening module communicated with the data filtering module (4) and the polarity check module (5)
(6) screening conditions that, the viewpoint information screening module (6) includes communicating preset unit (601), screening unit (602), row
Sequence unit (603), the screening conditions, which preset unit (601), is used for the default screening factor, the screening factor including the period,
At least one in follow-up number, thumb up number, the screening unit (602) is used to comment on interaction according to the screening factor pair
Some viewpoint informations in webpage are screened, and the viewpoint information filtered out is sent to the sequencing unit
(603), the sequencing unit (603) be used for will the viewpoint information that filter out according to the screening factor by high on earth suitable
Sequence is ranked up and sent to the polarity check module (5).
4. network social intercourse media viewpoint tendency analysis system as claimed in claim 1, it is characterised in that the polarity check
Module (5) includes the threshold preset unit (501) communicated, word extraction unit (502), matching degree judging unit (503), feelings
Feel tendentiousness processing unit (504), emotional orientation analysis unit (505), label for labelling unit (506)；
The threshold preset unit (501), which is used to preset the emotionality word or word in the positive emotion cluster (101), to be inclined
To threshold value+F, while for being inclined to threshold value-F to the emotionality word or word in the negative sense emotion cluster (102) are default, its
In, F is integer；
The word extraction unit (502) is used to extract some Sentiment orientation sex factors in the viewpoint information；
The matching degree judging unit (503) includes judgment sub-unit, retrieval subelement, separately deposits subelement, and judgement is single
Member is used in the Sentiment orientation sex factor of extraction and the positive emotion cluster (101) and the negative sense emotion cluster (102)
Emotionality word or word carry out matching degree judgement, when Sentiment orientation sex factor belongs to positive emotion cluster (101) i.e.
Belong to positive emotion word or word, belong to negative sense feelings when Sentiment orientation sex factor belongs to negative sense emotion cluster (102)
Feel word or word, the negative sense emotion is also not belonging to when Sentiment orientation sex factor neither belongs to the positive emotion cluster (101)
During cluster (102), then the Sentiment orientation sex factor is sent to the retrieval subelement, the retrieval subelement passes through retrieval
Sentiment orientation information of the Sentiment orientation sex factor in the historical events of the network social intercourse media, and send to described another
Subelement is deposited, the another subelement of depositing is used to according to its Sentiment orientation information correspondence preserve the Sentiment orientation sex factor to institute
State in positive emotion cluster (101) or the negative sense emotion cluster (102)；
Some emotion tendency factors difference that the emotion tendency processing unit (504) is used to count in the viewpoint information
Belong to the quantity of the positive emotion cluster (101) and the negative sense emotion cluster (102), and calculate described by below equation
The emotion tendency value W of viewpoint information：
W=N × (+F)+M × (- F)；
Wherein, W is the emotion tendency value W of the viewpoint information；N belongs to the positive emotion for the Sentiment orientation sex factor
The quantity of cluster (101)；M is the quantity that the Sentiment orientation sex factor belongs to the negative sense emotion cluster (102), and F is tendency
Threshold value；
The emotional orientation analysis unit (505) is used to analyze the viewpoint information according to the emotion tendency value W
Emotion tendency, is positive emotion when the emotion tendency value W is more than or equal to tendency threshold value+F；When the emotion is inclined
When tropism value W is less than or equal to tendency threshold value-F, as negative sense emotion；It is neutrality when the emotion tendency value W is equal to 0
Emotion；
The label for labelling unit (506) is used for the viewpoint letter analyzed to the emotional orientation analysis unit (505)
Breath assigns polarity label, and the polarity label includes positive emotion information, negative sense emotion information and neutral emotion information.
5. network social intercourse media viewpoint tendency analysis system as claimed in claim 1, it is characterised in that the analysis system
Also include viewpoint polarity statistical module (7) and the result display module (8) communicated with the polarity check module (5), it is described
Viewpoint polarity statistical module (7) is used to be united respectively according to the polarity label of some viewpoint informations in the comment Interactive Web Page
The quantity of the positive emotion information, the negative sense emotion information and the neutral emotion information is counted out, and is sent to the knot
Fruit display module (8)；The result display module (8) is used for the positive emotion information, the negative sense emotion information and institute
The quantity for stating neutral emotion information is drawn viewpoint analysis block diagram and sent to the database (1) preservation.
6. network social intercourse media viewpoint tendency analysis system as claimed in claim 1, it is characterised in that the webpage capture
Module (2) includes network address acquiring unit (201), the webpage capture unit (202), the network address acquiring unit (201) communicated
URL network address for obtaining the network social intercourse media, the webpage capture unit (202) is used to utilize spiders grabber
Capture the comment Interactive Web Page of the network social intercourse media.
7. a kind of network social intercourse media viewpoint sentiment classification method, it is characterised in that the analysis method comprises the following steps：
S1, by spiders the comment Interactive Web Page of network social intercourse media is captured, the network social intercourse media include
Microblogging, wechat, blog, forum, blog, transaction platform；
S2, some viewpoint informations for reading user comment in the comment Interactive Web Page；
S3, the viewpoint information screened, and by the viewpoint unrelated with commenting on theme in the network social intercourse media
Information deletion；
S4, some Sentiment orientation sex factors extracted in the viewpoint information, the Sentiment orientation sex factor include Sentiment orientation
Property word or emotion tendency word, and by the Sentiment orientation sex factor and the positive emotion cluster (101) and the negative sense feelings
The emotionality word or word felt in cluster (102) carry out matching degree judgement, while counting some emotions in the viewpoint information
Tendency Factor is belonging respectively to the quantity of the positive emotion cluster (101) and the negative sense emotion cluster (102), passes through quantity
Comparative analysis goes out the emotion tendency of the viewpoint information, while assigning polarity label, the polarity mark to the viewpoint information
Label include positive emotion information and negative sense emotion information.
8. network social intercourse media viewpoint sentiment classification method as claimed in claim 7, it is characterised in that right in step S3
The viewpoint information is screened, and the viewpoint information unrelated with commenting on theme in the network social intercourse media is deleted,
Specific method is：
Some key factors related to the comment theme in S3-1, the extraction comment Interactive Web Page；
S3-2, some key factors are trained as training sample to convolutional neural networks model, set up related
Spend model；
S3-3, the keyword extracted in the viewpoint information；
S3-4, the keyword inputted into the relevance model be trained, and draw the viewpoint information with it is described
Comment on the similarity output valve of theme；
S3-5, the viewpoint information by the similarity output valve less than default similarity threshold values are deleted.
9. network social intercourse media viewpoint sentiment classification method as claimed in claim 7, it is characterised in that the step S4 tools
Body includes following methods：
S4-1, tendency threshold value+F default to the emotionality word or word in the positive emotion cluster (101), while for institute
The default tendency threshold value-F of emotionality word or word in negative sense emotion cluster (102) is stated, wherein, F is integer；
S4-2, some Sentiment orientation sex factors extracted in the viewpoint information；；
S4-3, by the Sentiment orientation sex factor of extraction and the positive emotion cluster (101) and the negative sense emotion cluster (102)
Interior emotionality word or word carry out matching degree judgement, when Sentiment orientation sex factor belongs to positive emotion cluster (101)
Belong to positive emotion word or word, belong to negative sense when Sentiment orientation sex factor belongs to negative sense emotion cluster (102)
Emotion word or word；
Some emotion tendency factors in S4-4, the statistics viewpoint information are belonging respectively to the positive emotion cluster (101)
With the quantity of the negative sense emotion cluster (102), and the emotion tendency value W of the viewpoint information is calculated by below equation：
W=N × (+F)+M × (- F)；
Wherein, W is the emotion tendency value W of the viewpoint information；N belongs to the positive emotion for the Sentiment orientation sex factor
The quantity of cluster (101)；M is the quantity that the Sentiment orientation sex factor belongs to the negative sense emotion cluster (102), and F is tendency
Threshold value；
S4-5, the emotion tendency for analyzing according to the emotion tendency value W viewpoint information, when the emotion tendency
It is positive emotion when value W is more than or equal to tendency threshold value+F；When the emotion tendency value W is less than or equal to tendency threshold value-F,
As negative sense emotion；It is neutral emotion when the emotion tendency value W is equal to 0；
S4-6, the viewpoint information imparting polarity label to analyzing, the polarity label include positive emotion information, negative sense
Emotion information and neutral emotion information.
10. network social intercourse media viewpoint sentiment classification method as claimed in claim 7, it is characterised in that step S3 is also wrapped
Include and the viewpoint information is screened, screening technique is：
1. the screening factor is preset, the screening factor includes at least one in period, follow-up number, thumb up number；
2. some viewpoint informations in the comment Interactive Web Page according to the screening factor pair are screened；
3. the viewpoint information filtered out is ranked up according to the screening factor by high order on earth.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710160543.3A CN106951409A (en) | 2017-03-17 | 2017-03-17 | A kind of network social intercourse media viewpoint tendency analysis system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710160543.3A CN106951409A (en) | 2017-03-17 | 2017-03-17 | A kind of network social intercourse media viewpoint tendency analysis system and method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106951409A true CN106951409A (en) | 2017-07-14 |
Family
ID=59473841
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710160543.3A Pending CN106951409A (en) | 2017-03-17 | 2017-03-17 | A kind of network social intercourse media viewpoint tendency analysis system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106951409A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108470046A (en) * | 2018-03-07 | 2018-08-31 | 中国科学院自动化研究所 | Media event sort method and system based on media event search statement |
| CN108595564A (en) * | 2018-04-13 | 2018-09-28 | 众安信息技术服务有限公司 | Media friendliness appraisal procedure, device and computer readable storage medium |
| CN109241402A (en) * | 2018-07-31 | 2019-01-18 | 成都华栖云科技有限公司 | A kind of virtual comment machine introduction method based on news content |
| CN109446404A (en) * | 2018-08-30 | 2019-03-08 | 中国电子进出口有限公司 | A kind of the feeling polarities analysis method and device of network public-opinion |
| CN109740156A (en) * | 2018-12-28 | 2019-05-10 | 北京金山安全软件有限公司 | Feedback information processing method and device, electronic equipment and storage medium |
| CN110674415A (en) * | 2019-09-20 | 2020-01-10 | 北京浪潮数据技术有限公司 | Information display method and device and server |
| CN110992214A (en) * | 2019-11-29 | 2020-04-10 | 成都中科大旗软件股份有限公司 | Service management system and method based on tourist name county and demonstration area |
| CN112559743A (en) * | 2020-12-09 | 2021-03-26 | 深圳市网联安瑞网络科技有限公司 | Method, device, equipment and storage medium for calculating support degree of government and enterprise network |
| CN112966173A (en) * | 2019-12-13 | 2021-06-15 | 北京达佳互联信息技术有限公司 | Classification operation method and device for information comments |
| CN113177164A (en) * | 2021-05-13 | 2021-07-27 | 聂佼颖 | Multi-platform collaborative new media content monitoring and management system based on big data |
| CN113220823A (en) * | 2020-01-21 | 2021-08-06 | 北京中科闻歌科技股份有限公司 | Sentiment, topic and viewpoint analysis method for social media public language |
| CN114398473A (en) * | 2022-01-19 | 2022-04-26 | 平安国际智慧城市科技股份有限公司 | Enterprise portrait generation method, device, server and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102708096A (en) * | 2012-05-29 | 2012-10-03 | 代松 | Network intelligence public sentiment monitoring system based on semantics and work method thereof |
| DE102013000611A1 (en) * | 2013-01-16 | 2014-07-17 | i-market GmbH | Automatic method for recognizing brochures, catalogs or prospectus on websites of organizations, involves detecting and storing source code of to-be examined website by crawler or selecting source code completely or partially from database |
-
2017
- 2017-03-17 CN CN201710160543.3A patent/CN106951409A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102708096A (en) * | 2012-05-29 | 2012-10-03 | 代松 | Network intelligence public sentiment monitoring system based on semantics and work method thereof |
| DE102013000611A1 (en) * | 2013-01-16 | 2014-07-17 | i-market GmbH | Automatic method for recognizing brochures, catalogs or prospectus on websites of organizations, involves detecting and storing source code of to-be examined website by crawler or selecting source code completely or partially from database |
Non-Patent Citations (1)
| Title |
|---|
| 王君泽: "《网络舆情应对的关键技术研究》", 31 January 2017, 华中科技大学出版社 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108470046B (en) * | 2018-03-07 | 2020-12-01 | 中国科学院自动化研究所 | News event sequencing method and system based on news event search sentence |
| CN108470046A (en) * | 2018-03-07 | 2018-08-31 | 中国科学院自动化研究所 | Media event sort method and system based on media event search statement |
| CN108595564A (en) * | 2018-04-13 | 2018-09-28 | 众安信息技术服务有限公司 | Media friendliness appraisal procedure, device and computer readable storage medium |
| CN109241402A (en) * | 2018-07-31 | 2019-01-18 | 成都华栖云科技有限公司 | A kind of virtual comment machine introduction method based on news content |
| CN109446404A (en) * | 2018-08-30 | 2019-03-08 | 中国电子进出口有限公司 | A kind of the feeling polarities analysis method and device of network public-opinion |
| CN109446404B (en) * | 2018-08-30 | 2022-04-08 | 中国电子进出口有限公司 | Method and device for analyzing emotion polarity of network public sentiment |
| CN109740156A (en) * | 2018-12-28 | 2019-05-10 | 北京金山安全软件有限公司 | Feedback information processing method and device, electronic equipment and storage medium |
| CN109740156B (en) * | 2018-12-28 | 2023-08-04 | 北京金山安全软件有限公司 | Feedback information processing method and device, electronic equipment and storage medium |
| CN110674415A (en) * | 2019-09-20 | 2020-01-10 | 北京浪潮数据技术有限公司 | Information display method and device and server |
| CN110674415B (en) * | 2019-09-20 | 2022-06-17 | 北京浪潮数据技术有限公司 | Information display method and device and server |
| CN110992214A (en) * | 2019-11-29 | 2020-04-10 | 成都中科大旗软件股份有限公司 | Service management system and method based on tourist name county and demonstration area |
| CN112966173A (en) * | 2019-12-13 | 2021-06-15 | 北京达佳互联信息技术有限公司 | Classification operation method and device for information comments |
| CN112966173B (en) * | 2019-12-13 | 2024-02-27 | 北京达佳互联信息技术有限公司 | Classification operation method and device for information comments |
| CN113220823A (en) * | 2020-01-21 | 2021-08-06 | 北京中科闻歌科技股份有限公司 | Sentiment, topic and viewpoint analysis method for social media public language |
| CN113220823B (en) * | 2020-01-21 | 2024-03-01 | 北京中科闻歌科技股份有限公司 | Method and device for analyzing emotion, topic and viewpoint of social media public language |
| CN112559743A (en) * | 2020-12-09 | 2021-03-26 | 深圳市网联安瑞网络科技有限公司 | Method, device, equipment and storage medium for calculating support degree of government and enterprise network |
| CN112559743B (en) * | 2020-12-09 | 2024-02-13 | 深圳市网联安瑞网络科技有限公司 | Method, device, equipment and storage medium for calculating government and enterprise network support |
| CN113177164A (en) * | 2021-05-13 | 2021-07-27 | 聂佼颖 | Multi-platform collaborative new media content monitoring and management system based on big data |
| CN114398473A (en) * | 2022-01-19 | 2022-04-26 | 平安国际智慧城市科技股份有限公司 | Enterprise portrait generation method, device, server and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106951409A (en) | A kind of network social intercourse media viewpoint tendency analysis system and method | |
| CN109492157B (en) | News recommendation method and theme characterization method based on RNN and attention mechanism | |
| CN105740228B (en) | A kind of internet public feelings analysis method and system | |
| CN101820366B (en) | Pre-fetching-based fishing web page detection method | |
| CN103823844B (en) | Question forwarding system and question forwarding method on the basis of subjective and objective context and in community question-and-answer service | |
| CN109829089A (en) | Social network user method for detecting abnormality and system based on association map | |
| CN107577759A (en) | User comment auto recommending method | |
| CN110705288A (en) | Big data-based public opinion analysis system | |
| CN103425799A (en) | Personalized research direction recommending system and method based on themes | |
| Scrivens et al. | Searching for extremist content online using the dark crawler and sentiment analysis | |
| CN104809108A (en) | Information monitoring and analyzing system | |
| CN104834739B (en) | Internet information storage system | |
| CN104809252A (en) | Internet data extraction system | |
| CN115033668B (en) | Story venation construction method and device, electronic equipment and storage medium | |
| Safarnejad et al. | A multiple feature category data mining and machine learning approach to characterize and detect health misinformation on social media | |
| Daouadi et al. | Organization vs. Individual: Twitter User Classification. | |
| CN112417267A (en) | User behavior analysis method and device, computer equipment and storage medium | |
| Atoum | Detecting cyberbullying from tweets through machine learning techniques with sentiment analysis | |
| Dagar et al. | Twitter sentiment analysis using supervised machine learning techniques | |
| Spitters et al. | Threat detection in tweets with trigger patterns and contextual cues | |
| CN116723005A (en) | Method and system for tracking malicious code implicit information under polymorphic hiding | |
| CN113868536B (en) | Information recommendation method, device, equipment and storage medium | |
| Washha et al. | Information quality in social networks: Predicting spammy naming patterns for retrieving twitter spam accounts | |
| CN107609094A (en) | Data disambiguation method, device and computer equipment | |
| Liu et al. | Oasis: online analytic system for incivility detection and sentiment classification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170714 |