CN107305545A - A kind of recognition methods of the network opinion leader based on text tendency analysis - Google Patents

A kind of recognition methods of the network opinion leader based on text tendency analysis Download PDF

Info

Publication number
CN107305545A
CN107305545A CN201610240853.1A CN201610240853A CN107305545A CN 107305545 A CN107305545 A CN 107305545A CN 201610240853 A CN201610240853 A CN 201610240853A CN 107305545 A CN107305545 A CN 107305545A
Authority
CN
China
Prior art keywords
opinion
leader
comment
network
dictionary
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
Application number
CN201610240853.1A
Other languages
Chinese (zh)
Inventor
陈芬
彭玥
许青青
汤丽萍
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University of Science and Technology
Original Assignee
Nanjing University of Science and Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nanjing University of Science and Technology filed Critical Nanjing University of Science and Technology
Priority to CN201610240853.1A priority Critical patent/CN107305545A/en
Publication of CN107305545A publication Critical patent/CN107305545A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/951—Indexing; Web crawling techniques
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Databases & Information Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Machine Translation (AREA)

Abstract

The invention discloses a kind of method of the leader of opinion recognized based on text tendency analysis in network public-opinion.This method is specifically included：One is the index system that network opinion leader identification is built using modification method, identifies potential leader of opinion.Two be to add text tendency analysis, introduces Word2Vec algorithms, rejects negative emotion ratio overweight " pseudo- leader of opinion ".Three be the contrast verification of effect, and the leader of opinion that will identify that is contrasted with WeiboRank leader of opinion's algorithms, so as to verify the validity and confidence level of algorithm proposed by the present invention.The present invention can recognize that the leader of opinion of three quasi-representatives, cover each stage for being constantly expanded to the transformation of public opinion attitude and stabilization in public sentiment evolution from accident origin, influence power, to realize monitoring network public sentiment evolution process, catching and predicting that the people's livelihood will of the people provides theoretical foundation, and the important measures that prevention burst group accidents into practice can be used as to occur.

Description

A kind of recognition methods of the network opinion leader based on text tendency analysis
Technical field
The present invention relates to a kind of recognition methods, and in particular to a kind of network opinion leader based on text tendency analysis Recognition methods.
Background technology
With the development of internet, the popularity rate of social network sites is also improved constantly, it is this using microblogging, forum as representative net Network interpersonal communication mode has penetrated into the life of people gradually, as the masses for social phenomenon and social concern expression conviction, One of Important Platform of attitude, opinion and mood.Expression, propagation and the interaction of the condition of the people will of the people are known as network carriage on this line Feelings.Due to the spontaneity and freedom of network, the speech of rationality had both been included in network public-opinion, can also there is extreme speech even ballad Speech, so needing to take certain measure to be monitored early warning to network public-opinion.During network opinion leader is exactly such event, energy The VIP for enough helping the survival of public opinion subject under discussion and promoting public opinion to change, not only gives the discussion framework of topic, also invisible It is middle to influence the attitude of other netizens.Therefore by recognizing the network opinion leader in different event, it can quickly find netizen's Universal attitude and public opinion trend, are caught and are predicted to the people's livelihood will of the people with this, and can be used as prevention Derived Populations accident The important measures of generation.
It is existing that web page browsing, comment collection, viewpoint analysis can be carried out again by artificial to leader of opinion's knowledge method for distinguishing Go to recognize leader of opinion, but not only operating efficiency is low but also artificial evaluation criterion is inconsistent for this method, it is difficult to the online sea of reply The collection and processing of information are measured, it is necessary to strengthen the research of related information technology, a set of automated network leader of opinion is formed and knows Other system.And existing automation leader of opinion identifying system is main by Social Network Analysis Method ＆ and clustering method, this Class method focuses primarily upon to bloger's personal information and commented on the extraction and utilization of forwarding relation, has lacked public's comment attitude Recognition mechanism, although this may cause in the leader of opinion by screening occur can trigger netizen's extensive discussions, discuss Content is all the bloger for opposing or even abusing sound.These " pseudo- leaders of opinion " and the definition for being unsatisfactory for leader of opinion, can not yet Realize the basic role of leader of opinion.
The content of the invention
Text tendency analysis is carried out to comment language material it is an object of the invention to provide one kind, excavates and is full of emotional attitude Social event in online friend attitude so that really recognize network opinion leader method.
The technical solution for realizing the object of the invention is：A kind of network opinion leader based on text tendency analysis Recognition methods, step is as follows：
The first step, index system are set up, i.e., set up index by analyzing the different characteristic of leader of opinion first；Then utilize Analytic hierarchy process (AHP) calculates the corresponding weight of each index；The real data captured again by network is matched with index, finally Substitute into formula and obtain leader's value, potential leader of opinion is recognized by the ranking of fraction.
Second step, the extraction of evaluation object, i.e., by Stanford syntactic analysis methods, the syntactic structure to comment is carried out Dissect, so that it is, for the comment of bloger or for content, to realize the extraction of evaluation object to distinguish comment under microblogging.
3rd step, text tendency analysis is Text Pretreatment first, different language materials is made pauses in reading unpunctuated ancient writings in advance, lattice Formula processing, participle and part-of-speech tagging；Then dependence is extracted, i.e., syntactic analysis is carried out on the basis of subordinate sentence, sentence is found out In dependence and main word and qualifier；Dictionary is determined again, realizes that network sentiment neologisms are sent out using Word2vec models It is existing, positive and negative sentiment dictionary is improved, positive and negative dictionary, degree rank dictionary, negative word dictionary and punctuation mark dictionary is finally given.Most The calculating of laggard market sense fraction, main word is compared in positive and negative face dictionary and obtains initial word polarity, then by qualifier with Word degree rank dictionary and negative word dictionary, which are compared, obtains qualifier weight, the two multiplication is drawn the feelings of dependence level Feel fraction.Then the weights that the punctuation mark in sentence and sentence order are carried are extracted, by itself and all interdependent passes in sentence It is that emotion fraction sum is multiplied, it is possible to draw the Sentiment orientation fraction of this sentence.
4th step, recognizes leader of opinion, i.e., carries out above three as experiment language material to the data in crawl microblog The operation of step, identifies real network opinion leader, and carry out contrast verification with WeiboRank leaders of opinion algorithm.
The present invention is compared with the prior art, and its remarkable advantage includes：(1) structure of leader of opinion's distinguishing indexes system, one Aspect overall merit user in itself with the information such as customer relationship, on the other hand set up New Set --- " Media Exposure degree " and " OK Industry " extracts the comment forwarding relation between user, the items of network opinion leader is weighed comprehensively as the important evidence of identification Standard.(2) add text trend analysis method and semantic analysis is carried out to comment content, in the base of existing opinion leader identification method Increase netizen's comment attitude identification process on plinth, perfect leader of opinion recognizes the technology path of system, and rejecting can not represent group Many " pseudo- leaders of opinion ", makes recognition result more accurate credible.(3) using Google Word2Vec algorithms, one is formed More perfect sentiment dictionary.This method utilizes neural network model, and speed is fast and effect is good, is accurate analysis Text Orientation Lay good basis.
The present invention is described in further detail below in conjunction with the accompanying drawings.
Brief description of the drawings
Fig. 1 is the network opinion leader identification method flow chart of the invention based on text tendency analysis.
Fig. 2 is the index system of the network opinion leader identification built.
Fig. 3 is the flow chart of evaluation object screening.
Fig. 4 is the flow chart of text tendency analysis.
Embodiment
With reference to Fig. 1, the recognition methods of the network opinion leader of the invention based on text tendency analysis, step is as follows：
The first step, index system are set up：Index is set up by analyzing the different characteristic of leader of opinion first, as shown in Figure 2； Then the corresponding weight of each index is calculated using analytic hierarchy process (AHP)；The real data captured again by network is matched somebody with somebody with index It is right, finally substitute into formula and obtain leader's value, potential leader of opinion is recognized by the ranking of fraction.
Second step, the extraction of evaluation object：As shown in figure 3, by Stanford syntactic analysis methods, to the grammer of comment Structure is dissected, so that it is, for the comment of bloger or for content, to realize evaluation object to distinguish comment under microblogging Extract.
3rd step, text tendency analysis, as shown in figure 4, being Text Pretreatment first, different language materials is carried out in advance Punctuate, format analysis processing, participle and part-of-speech tagging；Then dependence is extracted, i.e., syntactic analysis is carried out on the basis of subordinate sentence, is looked for The dependence and main word and qualifier gone out in sentence；Dictionary is determined again, and network sentiment is realized using Word2vec models New word discovery, improves positive and negative sentiment dictionary, finally gives positive and negative dictionary, degree rank dictionary, negative word dictionary and punctuation mark Dictionary.The calculating of emotion fraction is finally carried out, main word is compared in positive and negative face dictionary and obtains initial word polarity, then will Qualifier is compared with word degree rank dictionary and negative word dictionary and obtains qualifier weight, and the two multiplication is drawn dependence The emotion fraction of level.Then the weights that the punctuation mark in sentence and sentence order are carried are extracted, by itself and institute in sentence There is the multiplication of dependence emotion fraction sum, it is possible to draw the Sentiment orientation fraction of this sentence.
4th step, recognizes leader of opinion, i.e., carries out above three as experiment language material to the data in crawl microblog The operation of step, identifies real network opinion leader, and carry out contrast verification with WeiboRank leaders of opinion algorithm.Under Face is specifically described with reference to example：
The present invention chooses " so-and-so " event in 40 points or so the generations of on 2 25th, 2014 morning as analysis object, first First capture the related microblogging of the event on Sina weibo platform, the time interval of crawl task on 2 25th, 2,014 0 point arrive 23 points of April 15 day in 2014.For the topic capture altogether 33079 original microbloggings, 424898 forwarding, 304750 comment and The essential information of 360549 microblog users, extracts 410418 forwarding relations altogether.Network based on text tendency analysis The recognition methods of leader of opinion includes herein below：
1st, preliminary leader of opinion's screening is carried out using modification method
The microblog data come to capturing obtains leader of opinion's specific targets data after arranging, and the dimension between data is carried out It is unified, obtain normalized data.After calculating, system sorts leader's value according to every bloger from high to low, this hair It is bright extract in all blogers before ranking 1% namely preceding 15 as potential leader of opinion, it is specific as shown in table 1.
Leader of opinion in table 1 " so-and-so " event
Bloger's pet name Leader is worth
Online friend A 0.829111545
Online friend B 0.651347868
Online friend C 0.461397104
Online friend D 0.412328209
Online friend E 0.367892438
Online friend F 0.363330525
Online friend G 0.32739149
Online friend H 0.310087118
Online friend I 0.287684557
Online friend J 0.267977784
Online friend K 0.245805917
Online friend L 0.244582583
Online friend M 0.219240969
Online friend N 0.218534647
Online friend O 0.216970145
2nd, evaluation object is screened
The comment under the bloger Jing Guo network opinion leader preliminary screening is saved into respective document respectively first, so Afterwards syntactic analysis is carried out using Stanford parser.After the parser of Stamford, the grammer of sentence can be obtained Tree, finds out subject therein and object.Then comment is screened using self-built dictionary, it is possible to obtain which comment is " comment for being directed to bloger ", which is " comment for being directed to content of microblog ".After evaluation object is screened, each possible opinion The comment number of leader is as shown in table 2.Because online friend H closes comment function, so it is 0 that it, which comments on number,.
The evaluation object of table 2 extracts result
Bloger's pet name All comment numbers For the comment number of bloger For the comment number of content
Online friend A 12320 112 11208
Online friend B 7689 311 7378
Online friend C 7862 162 7700
Online friend D 3206 5 3201
Online friend E 1378 66 1312
Online friend F 1651 121 1530
Online friend G 2371 54 2317
Online friend H 0 0 0
Online friend I 1577 18 1559
Online friend J 1584 833 825
Online friend K 1048 19 1029
Online friend L 2833 86 2747
Online friend M 284 16 268
Online friend N 66 9 57
Online friend O 120 19 101
3rd, text tendency analysis
The text tendency analysis of the present invention is divided into two parts on the basis of evaluation object screening：" for bloger's Comment " analysis and " comment for being directed to content of microblog " analysis.The comment of bloger " be directed to " by set forth herein Text Orientation Analysis system, is divided into 3 classes：" front ", " negative " and " neutrality ", represents to hold bloger approval attitude respectively, opposes attitude, do not have There is obvious tendentiousness." comment for being directed to content of microblog " is also divided into " just by text tendency analysis system Face ", " negative " and " neutrality ", but according to the difference of content of microblog, they have different meanings, it is specific as shown in table 3.
Table 3 comments on emotion tendency corresponding table for content of microblog
After the calculating of Text Orientation algorithm, obtain each potential microblogging bloger comment " front ", " negative " and " in It is vertical " quantity, the percentage of opposition is counted, help system rejecting " pseudo- leader of opinion " refers to table 4 and table 5.
Table 4 is directed to the text tendency analysis result that bloger comments on
Bloger's pet name Front It is neutral Negatively Opposition rate
Online friend A 17 74 21 0.188
Online friend B 39 204 68 0.219
Online friend C 19 78 65 0.401
Online friend D 0 2 3 0.600
Online friend E 30 19 17 0.258
Online friend F 12 81 28 0.231
Online friend G 6 30 18 0.333
Online friend H 0 0 0 -
Online friend I 2 4 2 0.250
Online friend J 115 530 187 0.225
Online friend K 6 2 11 0.579
Online friend L 20 49 17 0.198
Online friend M 3 10 3 0.188
Online friend N 2 6 1 0.111
Online friend O 5 7 7 0.368
Table 5 is directed to the text tendency analysis result that content of microblog is commented on
Bloger's pet name Front It is neutral Negatively Non-correlation
Online friend A 952 9321 935 0.085
Online friend B 701 5275 1402 0.095
Online friend C 693 6006 1001 0.090
Online friend D 560 1568 1072 0.175
Online friend E 131 984 197 0.100
Online friend F 176 1010 344 0.115
Online friend G 209 1552 556 0.090
Online friend H 0 0 0 -
Online friend I 130 1081 348 0.083
Online friend J 89 561 176 0.108
Online friend K 117 642 270 0.114
Online friend L 295 1727 725 0.107
Online friend M 31 191 46 0.116
Online friend N 8 45 4 0.140
Online friend O 14 76 11 0.139
4th, the validity check and contrast of leader of opinion's identification
Obtain after the Sentiment orientation of all comments, it is necessary to by the opposition rate commented on for bloger with being commented for content of microblog The uncorrelated rate of opinion is calculated in proportion, and formula is as follows：
Wherein xiFor the opposition rate commented on for bloger, n is " comment for being directed to bloger " quantity；yiFor for content of microblog The uncorrelated rate of comment, N is " comment for being directed to content of microblog " quantity, and final calculation result is shown in Table 6.
The bloger of table 6 " pseudo- leader of opinion " possibility result is counted
The present invention rejects the bloger of 3 before " pseudo- leader of opinion " possibility ranking the ranks of leader of opinion, because " online friend D ", " online friend H " and " comment under online friend's O " microbloggings is more be oppose bloger's opinion, do not believe that bloger's language or with comment content Unrelated.By manually consulting comment detection, " the comment content under online friend's J " relevant microblogs opposes and uncorrelated rate reaches 86% More than, " online friend D " is more than 59%, and " online friend O " is more than 65%, it was demonstrated that " the pseudo- leader of opinion " identified herein does not have really Obtain online friend's support, it is impossible to be designated as real network opinion leader.In addition, " online friend H " is but closed as one of popular bloger Comment function has been closed, can not play a part of that leader of opinion is due to widen one's influence, guide public opinion, therefore has also been deleted Remove.11 blogers finally left are only the leader of opinion in Sina weibo " so-and-so " event, refer to table 7.
The leader of opinion that the present invention of table 7 is identified
Sequence number Bloger's pet name
1 Online friend A
2 Online friend B
3 Online friend C
4 Online friend E
5 Online friend F
6 Online friend G
7 Online friend I
8 Online friend K
9 Online friend M
10 Online friend N
11 Online friend O
By the manual verification to 11 leaders of opinion in table 8, this 11 leaders of opinion really can be microblogging " so-and-so " Play vital effect in event, and these leaders of opinion can be summarized as event mode leader of opinion (such as event is witnessed Person " online friend H "), colony type leader of opinion (such as possess a large amount of beans vermicelli official's microblogging " online friend A " and " bloger online friend B "), Viewpoint type leader of opinion (such as delivered the viewpoint of oneself and approved of extensively " online friend F " and " online friend M " blogers).
In order to contrast the validity and reliability of recognition methods, by the algorithm of the present invention and based on PageRank's The leader of opinion that WeiboRank algorithms are drawn under same data set is contrasted.The opinion neck that WeiboRank algorithms are identified Sleeve is as shown in table 8.
The leader of opinion that table 8WeiboRank algorithms are identified
Ranking Microblog users name WR values
1 Online friend P 0.081777514
2 Online friend B 0.043827157
3 Online friend Q 0.029573971
4 Online friend D 0.027628687
5 Online friend E 0.026071973
6 Online friend R 0.022131157
7 Online friend K 0.02103858
8 Online friend I 0.01366443
9 Online friend S 0.011458707
10 Online friend G 0.010990369
11 Online friend A 0.010705298
12 Online friend T 0.010053468
13 Online friend U 0.008730804
14 Online friend V 0.008727109
15 Online friend J 0.008396438
From table 7 and table 8, experimental result of the invention has similar to WeiboRank algorithm acquired results, has 8 Leader of opinion is to overlap, and repetitive rate has reached more than 50%, thus demonstrates the inventive method in identification network opinion leader Validity on " transfer quality ".And be advantageous in that compared to WeiboRank methods can be more for the recognition methods of the present invention Sufficiently consider indices as leader of opinion, by leader's value more accurately calculate that bloger turns into leader of opinion can Energy property, can greatly improve the opinion finally identified with the pseudo- leader of opinion of exclusive PCR recognition result on this basis The validity of leader's result.

Claims (5)

1. a kind of method of the leader of opinion recognized based on text tendency analysis in network public-opinion, it is characterised in that step is such as Under：
Step 1, index system are set up, i.e., set up index by analyzing the different characteristic of leader of opinion first；Then level is utilized Analytic approach calculates the corresponding weight of each index；The real data captured again by network is matched with index, is finally substituted into Leader's value is obtained in formula, potential leader of opinion is recognized by the ranking of fraction；
The extraction of step 2, evaluation object, i.e., by Stanford syntactic analysis methods, the syntactic structure to comment is dissected, It is, for the comment of bloger or for content, to realize the extraction of evaluation object so as to distinguish comment under microblogging；
Step 3, text tendency analysis, i.e., be Text Pretreatment first, different language materials is made pauses in reading unpunctuated ancient writings in advance, at form Reason, participle and part-of-speech tagging；Then dependence is extracted, i.e., syntactic analysis is carried out on the basis of subordinate sentence, is found out in sentence Dependence and main word and qualifier；Dictionary is determined again, and network sentiment new word discovery is realized using Word2vec models, it is complete It is apt to positive and negative sentiment dictionary, finally gives positive and negative dictionary, degree rank dictionary, negative word dictionary and punctuation mark dictionary；It is most laggard The calculating of market sense fraction, main word is compared in positive and negative face dictionary and obtains initial word polarity, then by qualifier and word Degree rank dictionary and negative word dictionary, which are compared, obtains qualifier weight, the two multiplication is drawn the emotion point of dependence level Number；Then the weights that the punctuation mark in sentence and sentence order are carried are extracted, by itself and all dependence feelings in sentence Feel fraction sum to be multiplied, it is possible to draw the Sentiment orientation fraction of this sentence；
The data captured in microblog, i.e., carried out above three step by step 4, identification leader of opinion as experiment language material Operation, identify real network opinion leader, and contrast verification is carried out with WeiboRank leaders of opinion algorithm.
2. the side of leader of opinion recognized based on text tendency analysis in network public-opinion according to claim 1 a kind of Method, it is characterised in that in step 1, the process of Index Establishment is：All indexs in existing literature are integrated first, and by item All indexs are included in survey and carry out Analysis and Screening by " public's investigation of a networked society collaborative innovation research " of mesh group And reprocessing；It is " Media Exposure degree " and " professional " to add two New Sets in conjunction with expert opinion, constructs the present invention's Leader of opinion's distinguishing indexes system.
3. the side of leader of opinion recognized based on text tendency analysis in network public-opinion according to claim 1 a kind of Method, it is characterised in that in step 1, the corresponding weight of each index is calculated based on analytic hierarchy process (AHP), and calculating process is：Set up layer Leader of opinion's distinguishing indexes system is the hierarchical structure needed for analytic hierarchy process (AHP) in secondary structure, the present invention；With reference to expert opinion Multilevel iudge matrix between experience construction each two index；Calculate weight vector and do consistency check, obtain the group of each index Close weight.
4. the side of leader of opinion recognized based on text tendency analysis in network public-opinion according to claim 1 a kind of Method, it is characterised in that in step 2, by Stanford syntactic analysis methods, the syntactic structure to comment is analyzed, specifically Step is：Input comment content, Stanford syntactic analyses, extraction subject and object, comparison dictionary；If there is dictionary in comment In words, then the comment belongs to the comment of bloger " be directed to ", is otherwise considered as " comment for being directed to content of microblog ".
5. the side of leader of opinion recognized based on text tendency analysis in network public-opinion according to claim 1 a kind of Method, it is characterised in that in step 3, network sentiment new word discovery, Word2Vec algorithm conducts are realized using Word2vec models Word, using the semantic relation between word, can be converted into term vector, then utilize word by Google term vector instrument of increasing income Semantic distance relation between vector, automatic identification network sentiment neologisms, significantly improves the accuracy of Text Orientation.
CN201610240853.1A 2016-04-18 2016-04-18 A kind of recognition methods of the network opinion leader based on text tendency analysis Pending CN107305545A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610240853.1A CN107305545A (en) 2016-04-18 2016-04-18 A kind of recognition methods of the network opinion leader based on text tendency analysis

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610240853.1A CN107305545A (en) 2016-04-18 2016-04-18 A kind of recognition methods of the network opinion leader based on text tendency analysis

Publications (1)

Publication Number Publication Date
CN107305545A true CN107305545A (en) 2017-10-31

Family

ID=60152050

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610240853.1A Pending CN107305545A (en) 2016-04-18 2016-04-18 A kind of recognition methods of the network opinion leader based on text tendency analysis

Country Status (1)

Country Link
CN (1) CN107305545A (en)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI657395B (en) * 2018-02-09 2019-04-21 麟數據科技股份有限公司 Opinion leader related network-based trading system, method, and storage medium
CN109857871A (en) * 2019-01-28 2019-06-07 重庆邮电大学 A kind of customer relationship discovery method based on social networks magnanimity context data
CN110489658A (en) * 2019-07-12 2019-11-22 北京邮电大学 Online social network opinion leader method for digging based on digraph model
CN111027328A (en) * 2019-11-08 2020-04-17 广州坚和网络科技有限公司 Method for judging emotion positive and negative and emotional color of comments through corpus training
CN111159402A (en) * 2019-12-13 2020-05-15 深圳大学 Mining method of network user influence relationship based on topic opinion analysis and opinion leader
CN111177526A (en) * 2018-11-12 2020-05-19 百度在线网络技术（北京）有限公司 Network opinion leader identification method and device
CN112685621A (en) * 2021-01-06 2021-04-20 深圳市网联安瑞网络科技有限公司 Network public opinion detection system and method integrating public opinion wind direction tracking and civil opinion prediction
CN113111269A (en) * 2021-05-10 2021-07-13 网易（杭州）网络有限公司 Data processing method and device, computer readable storage medium and electronic equipment

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI657395B (en) * 2018-02-09 2019-04-21 麟數據科技股份有限公司 Opinion leader related network-based trading system, method, and storage medium
CN111177526A (en) * 2018-11-12 2020-05-19 百度在线网络技术（北京）有限公司 Network opinion leader identification method and device
CN111177526B (en) * 2018-11-12 2023-08-15 百度在线网络技术（北京）有限公司 Network opinion leader identification method and device
CN109857871A (en) * 2019-01-28 2019-06-07 重庆邮电大学 A kind of customer relationship discovery method based on social networks magnanimity context data
CN110489658A (en) * 2019-07-12 2019-11-22 北京邮电大学 Online social network opinion leader method for digging based on digraph model
CN111027328A (en) * 2019-11-08 2020-04-17 广州坚和网络科技有限公司 Method for judging emotion positive and negative and emotional color of comments through corpus training
CN111027328B (en) * 2019-11-08 2024-03-26 广州坚和网络科技有限公司 Method for judging comment emotion positive and negative and emotion color through corpus training
CN111159402A (en) * 2019-12-13 2020-05-15 深圳大学 Mining method of network user influence relationship based on topic opinion analysis and opinion leader
CN111159402B (en) * 2019-12-13 2023-06-30 深圳大学 Mining method for network user influence relation based on topic opinion analysis and opinion leader
CN112685621A (en) * 2021-01-06 2021-04-20 深圳市网联安瑞网络科技有限公司 Network public opinion detection system and method integrating public opinion wind direction tracking and civil opinion prediction
CN113111269A (en) * 2021-05-10 2021-07-13 网易（杭州）网络有限公司 Data processing method and device, computer readable storage medium and electronic equipment

Similar Documents

Publication Publication Date Title
CN107305545A (en) A kind of recognition methods of the network opinion leader based on text tendency analysis
CN101408883B (en) Method for collecting network public feelings viewpoint
CN105740228B (en) A kind of internet public feelings analysis method and system
CN103425799B (en) Individuation research direction commending system and recommend method based on theme
CN103500175B (en) A kind of method based on sentiment analysis on-line checking microblog hot event
Alsaedi et al. Arabic event detection in social media
CN113515634B (en) Social media rumor detection method and system based on hierarchical heterogeneous graph neural network
Liew et al. EmoTweet-28: A fine-grained emotion corpus for sentiment analysis
CN103729474B (en) Method and system for recognizing forum user vest account
Schulz et al. Crisis information management in the Web 3.0 age.
CN106940732A (en) A kind of doubtful waterborne troops towards microblogging finds method
CN106294532A (en) The image appraisal algorithm analyzed based on microblog emotional
CN103793503A (en) Opinion mining and classification method based on web texts
CN110990683B (en) Microblog rumor integrated identification method and device based on region and emotional characteristics
CN107291886A (en) A kind of microblog topic detecting method and system based on incremental clustering algorithm
CN103150333A (en) Opinion leader identification method in microblog media
CN105183717A (en) OSN user emotion analysis method based on random forest and user relationship
KR20120108095A (en) System for analyzing social data collected by communication network
CN104899335A (en) Method for performing sentiment classification on network public sentiment of information
CN105740382A (en) Aspect classification method for short comment texts
CN110297986A (en) A kind of Sentiment orientation analysis method of hot microblog topic
Yuan et al. A hybrid method for multi-class sentiment analysis of micro-blogs
CN114742071A (en) Chinese cross-language viewpoint object recognition and analysis method based on graph neural network
JP2009157450A (en) Mail sorting system, mail retrieving system, and mail destination sorting system
Zhao et al. Towards events detection from microblog messages

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20171031

WD01 Invention patent application deemed withdrawn after publication