CN106295702B - A kind of social platform user classification method based on the analysis of individual affective behavior - Google Patents
A kind of social platform user classification method based on the analysis of individual affective behavior Download PDFInfo
- Publication number
- CN106295702B CN106295702B CN201610668449.4A CN201610668449A CN106295702B CN 106295702 B CN106295702 B CN 106295702B CN 201610668449 A CN201610668449 A CN 201610668449A CN 106295702 B CN106295702 B CN 106295702B
- Authority
- CN
- China
- Prior art keywords
- user
- forwarding
- emotion
- individual
- tree
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Abstract
The invention discloses a kind of social platform user classification methods based on affective behavior analysis, comprising the following steps: S1, building forwarding tree；S2, building user's history record extract the individual forwarding history record that the node with same subscriber ID constructs this user；S3, building user feeling behavior description feature；S4, the user role disaggregated model that decision tree is given using the feature construction in S3 complete the social platform user analyzed based on affective behavior classification.The present invention constructs more comprehensive user feeling behavior description model, can more fully consider the case history information of user；This method takes full advantage of the userspersonal information in microblogging, transmission structure information, emotion information and dynamic time-domain information.Due to using the above measure, the present invention can obtain better classification accuracy.
Description
Technical field
The invention belongs to social networks technical field, in particular to a kind of social platform based on the analysis of individual affective behavior
User classification method.
Background technique
With the development of internet technology, it is used on a large scale by the online social networks of representative of microblogging.User
It can voluntarily release news on it, the modes such as can also thumb up by forwarding, comment and be interacted with other information, and it is true
Real social networks is identical, and what the user behavior of online social networks expressed is not only literal message, it includes simultaneously use
The emotional attitude at family, this emotional attitude is because that individual subscriber background is from the difference of habit is different, and through all of user
In interbehavior, we are referred to as this affective characteristics possessed by user the emotion role of user.
It mainly include at present the following aspects for the research of online social network user, 1, the digging of user force
Pick, such research are put forth effort on through the analysis to individual subscriber attribute and characteristics in spreading information, and description user social contact shadow is established
The model or algorithm of power are rung, realizes that user force calculates, finds social leader；2, the prediction of the online behavior of user, it is such
By to user's history, the considerations of factors such as context environmental and social networks, models user for research, realize to
The prediction of family specific behavior or preference, for example whether participating in forwarding, if interested etc..3, user feeling is analyzed, such research
What kind of emotion is had as starting point using some moment user, passes through multiple data sources (including text, picture, video, sound
It is happy etc.), the factors such as on-line off-line combination and social networks realize the analysis and prediction of user feeling.The above research is in certain journey
The online Behavior law of user and the inherent operation law of social networks are disclosed on degree for us, but is lacked to user feeling
Comprehensively consider.
Summary of the invention
In view of the above problems, the present invention is provided a kind of based on individual feelings by analyzing from individual subscriber emotion angle
Feel the social platform user classification method of behavioural analysis, the specific technical proposal is:
A kind of social platform user classification method based on the analysis of individual affective behavior, comprising the following steps:
S1, building forwarding tree: social platform user forwarding information is extracted, the social platform based on tree topology is established
Forwarding tree；
S2, building user's history record: affection computation is carried out for the forwarding information of the node in forwarding tree, result is pressed
Emotional semantic classification is positive, passive, neutral；Extract the individual forwarding history note that the node with same subscriber ID constructs this user
Record；
S3, building user feeling behavior description feature: Expressive Features are inclined to including user: individual and group's emotional relationship
ERu, individual subscriber history emotion preference HPu；User feeling influences Expressive Features EIu；
S4, the user role disaggregated model that decision tree is given using the feature construction in S3, first construction input vector Uu
=< ERu,HPu,EIu>, comentropy then is calculated separately to each feature
UjFor j-th of feature, the feature construction current decision node with maximum information gain is chosen, residue character step-by-step recursion is obtained
To final decision tree-model, and then complete user's classification based on emotion.
Further, the forwarding information in a kind of social platform user classification method S1 based on the analysis of individual affective behavior
Including urtext information, forwarding text information, participating user individual information.
Further, a kind of social platform user classification method S1 based on the analysis of individual affective behavior is according to level the bottom of by
Text emotion parsing is carried out upwards, successively adds forward node, constructs forwarding tree.
Further, the affection computation in a kind of social platform user classification method S2 based on the analysis of individual affective behavior
Using more rules collection model, sentiment dictionary, syntax rule are established by the way that text point mutual information is bottom-up, described is bottom-up
Refer to and is successively analyzed according to from the sequence of word, phrase, short sentence, whole sentence.
Further, individual described in a kind of social platform user classification method S3 based on the analysis of individual affective behavior
The distribution of emotion selection and group's emotion based on individual with group emotional relationship, be described as individual with when previous this letter of provision
The emotional relationship factor ER of breathu(w), value range is -1~1, and the bigger expression current relation of the value more approaches actively, the value
Smaller expression current relation more approaches passiveness, following to indicate:
Wherein, N (w), P (w), O (w) respectively indicate the distribution of the Negative Affect in current forwarding tree, and positive emotion is distributed, in
Vertical emotion distribution, S (w) indicate forwarding tree scale.
Further, individual history feelings in a kind of social platform user classification method S3 based on the analysis of individual affective behavior
Feel preference HPuIt (e) is based on the user comment participation C in the emotion distribution and history forwarding in user's history recordu(w),
It is indicated with following formula:
Wherein, exp {-θ1(t0-tw) it is the time decaying for controlling user preference, log (Cu(w) it+2) is grown to pass through to comment on
The degree of participation of degree description user.
Further, emotion described in a kind of social platform user classification method S3 based on the analysis of individual affective behavior
Influence EIuIt is the design feature SF based on forwarding treeu(w), the time domain of forwarding tree influences TFu(w), the emotion of user changes EIu
(w), following to indicate:
HRuIndicate that user forwards the number as internal node, NRuIndicate that user forwards the number as leaf node.
Further, it is a kind of based on individual affective behavior analysis social platform user classification method in forwarding tree structure
Feature SFu(w) the sheer size S (w) based on forwarding tree, relative size Su(w) and subtree depth DPu(w), following to indicate:
Further, it is a kind of based on individual affective behavior analysis social platform user classification method in forwarding tree time domain
Influence TFu(w) contribution propagated in time angle information for forwarding tree, the contribution are embodied in subtree relative to entire forwarding
Two aspects of time delay of the time-to-live, subtree of tree relative to urtext；
Wherein LPuIt (w) is subtree life cycle, LP (w) is forwarding tree life cycle,It is subtree relative to entire
The time-to-live of forwarding tree, exp {-ε (tu-tw) it is the time domain delay that subtree occurs；
Further, the emotion of user becomes in a kind of social platform user classification method based on the analysis of individual affective behavior
Change EFu(w) using the forwarding behavior of active user as time separation, the emotion of front and back is forwarded to be distributed by calculating user poor
It is different, and parameter is standardized by exponential function, it is indicated with following formula:
Wherein, Bu(w, e), Au(w, e) is respectively the emotion distribution that user forwards front and back.
The invention has the following advantages:
In order to the online affective behavior of description user of system, invention defines six class microblog users emotion roles,
It is positive leader respectively, positive follower, passive leader, passive follower, neutral leader, neutral follower, and mention
A kind of social platform user classification method based on the analysis of individual affective behavior out, this method from two dimensions (Sentiment orientation with
Emotion influences) establish user feeling behavior description model.
Due to using user feeling tendency feature and customer impact feature in technical solution, construct more comprehensive
User feeling behavior description model can more fully consider the case history information of user；This method takes full advantage of microblogging
Userspersonal information in the middle, transmission structure information, emotion information and dynamic time-domain information.Due to using the above measure, this
Invention can obtain better classification accuracy.
Detailed description of the invention
A kind of social platform user classification method flow chart based on the analysis of individual affective behavior of Fig. 1 present invention；
A kind of social platform user classification method user's history based on the analysis of individual affective behavior of Fig. 2 present invention records real
Example；
A kind of social platform user classification method architectural characteristic distribution based on the analysis of individual affective behavior of Fig. 3 present invention；
A kind of social platform user classification method time domain specification distribution based on the analysis of individual affective behavior of Fig. 4 present invention；
A kind of social platform user classification method parameter learning result based on the analysis of individual affective behavior of Fig. 5 present invention；
A kind of social platform user classification method emotion variation characteristic point based on the analysis of individual affective behavior of Fig. 6 present invention
Cloth；
A kind of social platform user classification method individual based on the analysis of individual affective behavior of Fig. 7 present invention and macroscopical emotion
Relationship distribution；
A kind of social platform user classification method history emotion preference knot based on the analysis of individual affective behavior of Fig. 8 present invention
Fruit distribution；
A kind of social platform user classification method emotion based on the analysis of individual affective behavior of Fig. 9 present invention influences result.
Specific embodiment
In order to which objects and advantages of the present invention are more clearly understood, the present invention is carried out with reference to embodiments further
It is described in detail.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not used to limit this hair
It is bright.
Embodiment
S1, building forwarding tree: social platform user forwarding information is extracted, the social platform based on tree topology is established
Forwarding tree
By taking microblogging as an example, the forwarding data on microblogging are grabbed, retain the user information in data, forwarding information and original
Beginning micro-blog information carries out text according to level according to the identifier " //@" and higher level user's pet name of microblogging forwarding from bottom to up
Forward node is successively added in parsing, constructs microblogging forwarding tree.It is collected into 19389 user informations in total, constructs forwarding tree 7096
?.
S2, building user's history record: affection computation is carried out for the forwarding information of the node in forwarding tree, result is pressed
Emotional semantic classification is positive, passive, neutral；Extract the individual forwarding history note that the node with same subscriber ID constructs this user.
Using more rules collection model, affection computation is carried out to the text information that each node is included in forwarding tree, is obtained
To three kinds as a result, being actively passive and neutrality respectively.Later, the user information for being included using each microblogging forward node,
The case history forwarding for constructing user will be come out with the Node extraction of same subscriber ID to record and deposit in the form of XML file
Storage.Fig. 2 is the historical record example of a user,<uid_1796678344>a user is represented,<retweet>currently to use
One forwarding at family,<org_id>,<org_text>,<org_time>,<org_emotion>,<p_name>,<p_id>,<w_
Id>,<w_test>,<w_time>,<w_emotion>indicate the association attributes of corresponding forwarding.
S3, building user feeling behavior description feature: Expressive Features are inclined to including user: individual and group's emotional relationship
ERu, individual subscriber history emotion preference HPu；User feeling influences Expressive Features EIu。
It is inclined to from personal with two angle building user feelings of macroscopical emotional relationship and individual subscriber history emotion preference,
For the former, with ERu(w) user is indicated and when the emotional relationship factor value range of previous microblogging is between -1~1, the value
Bigger expression current relation more approaches actively, otherwise approach is passive, to be located at neutral emotion near 0, setting actively with passiveness
Origin be 0.5 and -0.5 respectively,
N (w), P (w), O (w) respectively indicate the distribution of the three classes emotion in current forwarding tree (passive, actively, neutral), S (w)
Indicate forwarding tree scale.
Individual subscriber history emotion preference HPu(e) based in the emotion distribution and history forwarding in user's history record
User comment participation Cu(w), exponential part is used to control the time decaying of user preference, with nearest microblogging issuing time
t0As a reference point, logarithm part describes the degree of participation of user by commenting on length:
From the architectural characteristic of forwarding, time domain specification and emotion angle changing describe user feeling influence, microblogging forwarding
Design feature SFu(w) weigh sheer size S (w), the relative size S of forwarding treeu(w) and subtree depth DPu(w):
Fig. 3 describes SFu(w) calculated result distribution, it is believed that, with identical forwarding scale, son
Tree is deeper to mean that subtree is more sparse, on the contrary then denseer, and often there is denseer subtree larger range of influence to make
With.
Different from architectural characteristic, time domain influences TFu(w) it is used to describe the tribute that forwarding tree propagates information in time angle
It offers, in terms of this contribution embodies a concentrated reflection of two, first, subtree is relative to the time-to-live entirely forwarded；Second, subtree is opposite
In the time delay of original microblogging.TFu(w) comprehensively consider subtree life cycle LPu(w), forwarding tree life cycle LP (w) and
The time domain delay exp {-ε (t that subtree occursu-tw)}.ε figure is for controlling the rate of decay:
By experimental accuracy in this method, it is set to 0.2, Fig. 4 and describes the calculated result distribution of TFu (w).α1With
β1It is step-length with 0.1, selects the highest value of accuracy as ginseng by taking individually classification verifying to feature for learning parameter
Actual numerical value is counted, test results are shown in figure 5 using the classification method of decision tree in this reason, therefore parameter value is set to 0.6
With 0.7.
Emotion changes EFu(w) using the forwarding behavior of active user as time separation, user forwards the emotion point of front and back
Cloth is respectively with Bu(w, e), Au(w, e) is indicated, is passed through | Bu(w,e)-Au(w, e) | emotion distributional difference is calculated, and passes through index letter
Several pairs of parameters are standardized:
Fig. 6 describes EFu(w) calculated result distribution.
S4, the user role disaggregated model that decision tree is given using the feature construction in S3, first construction input vector Uu
=< ERu,HPu,EIu>, comentropy then is calculated separately to each feature
UjFor j-th of feature, the feature construction current decision node with maximum information gain is chosen, residue character step-by-step recursion is obtained
To final decision tree-model, and then complete user's classification based on emotion.
Fusion Features are carried out according to the result that S3 is obtained, obtain comprehensive description user feeling tendency ERu、HPuIt is influenced with emotion
EIuFeature as mode input:
Wherein EIuFeature is influenced on three classes to merge, and considers that leaf node does not generate any influence this case,
It introducesAs denoising factor HRuIndicate that user forwards the number as internal node, NRuIndicate user's forwarding
As the number of leaf node, Fig. 7 illustrates current data set ERuCalculated result distribution, Fig. 8 illustrates HPuCalculated result
Distribution, Fig. 9 illustrate EIuCalculated result distribution.Eventually by the classification method based on decision tree, 6 kinds of emotion roles are obtained
Classification, classification results are as shown in table 1.
1 embodiment emotion Role Classification result of table
| Emotion role | Accuracy |
| Positive leader (PL) | 0.87 |
| Positive follower (PF) | 0.90 |
| Neutral leader (OL) | 0.83 |
| Neutral follower (OF) | 0.86 |
| Passive leader (NL) | 0.91 |
| Passive follower (NF) | 0.92 |
Claims (7)
1. a kind of social platform user classification method based on the analysis of individual affective behavior, comprising the following steps:
S1, building forwarding tree: extracting social platform user forwarding information, establishes the social platform forwarding based on tree topology
Tree:
S2, building user's history record: affection computation is carried out for the forwarding information of the node in forwarding tree, result is pressed into emotion
It is classified as positive, passive, neutral: extracting the individual forwarding history record that the node with same subscriber ID constructs this user:
S3, building user feeling behavior description feature: Expressive Features are inclined to including user: individual and group's emotional relationship ERu, use
Family case history emotion preference HPu: user feeling influences Expressive Features EIu；
The described individual and group's emotional relationship are the distributions of emotion selection and group's emotion based on individual, be described as individual and
As the emotional relationship factor ER of previous text messageu(w), value range is -1~1, and the bigger expression current relation of the value is more
Actively, the smaller expression current relation of the value more approaches passiveness to approach, following to indicate:
Wherein, N (w), P (w), O (w) respectively indicate the distribution of the Negative Affect in current forwarding tree, positive emotion distribution, neutral feelings
Sense distribution, S (w) indicate forwarding tree scale；
The individual history emotion preference HPuIt (e) is based in the emotion distribution and history forwarding in user's history record
User comment participation Cu(w), it is indicated with following formula:
Wherein, exp [- θ1(t0-tw) it is the time decaying for controlling user preference, log (cu(w) it+2) is described to pass through comment length
The degree of participation of user；
It is the design feature SF based on forwarding tree that the emotion, which influences EIu,u(w), the time domain of forwarding tree influences TFu(w), user
Emotion change EFu(w), following to indicate:
Wherein, HRuIndicate that user forwards the number as interior N node, NRuIndicate that user forwards the number as leaf node；
S4, the user role disaggregated model that decision tree is given using the feature construction in S3, first construction input vector Uu=<
ERu, HPu, EIuThen > calculates separately comentropy to each featureUjFor
J-th of feature is chosen the feature construction current decision node with maximum information gain, is obtained most to residue character step-by-step recursion
Whole decision-tree model, and then complete user's classification based on emotion.
2. a kind of social platform user classification method based on the analysis of individual affective behavior according to claim 1, special
Sign is: the forwarding information in the S1 includes urtext information, the individual information for forwarding text information, participating user.
3. a kind of social platform user classification method based on the analysis of individual affective behavior according to claim 1, special
Sign is: the S1 carries out text emotion parsing according to level from bottom to up, successively adds forward node, constructs forwarding tree.
4. a kind of social platform user classification method based on the analysis of individual affective behavior according to claim 1, special
Sign is: the S2In affection computation use more rules collection model, establish emotion by the way that text point mutual information is bottom-up
Dictionary, syntax rule, it is described it is bottom-up refer to successively analyzed according to from the sequence of word, phrase, short sentence, whole sentence.
5. a kind of social platform user classification method based on the analysis of individual affective behavior according to claim 1, special
Sign is: the design feature SF of the forwarding treeu(w) the sheer size S (w) based on forwarding tree, relative size Su(w) and
Subtree depth DPu(w), following to indicate:
6. a kind of social platform user classification method based on the analysis of individual affective behavior according to claim 1, special
Sign is: the time domain of the forwarding tree influences TFu(w) contribution that information is propagated in time angle for forwarding tree, the contribution
It is embodied in two aspects of time delay of subtree relative to time-to-live of entire forwarding tree, subtree relative to urtext；
Wherein LPuIt (w) is subtree life cycle, LP (w) is forwarding tree life cycle,It is subtree relative to entire forwarding
The time-to-live of tree, exp {-e (tu-tw) it is the time domain delay that subtree occurs.
7. a kind of social platform user classification method based on the analysis of individual affective behavior according to claim 1, special
Sign is: the emotion of the user changes EFu(w) using the forwarding behavior of active user as time separation, pass through calculating
User forwards the emotion distributional difference of front and back, and is standardized by exponential function to parameter, is indicated with following formula:
Wherein, Bu(w, e), Au(w, e) is respectively the emotion distribution that user forwards front and back.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610668449.4A CN106295702B (en) | 2016-08-15 | 2016-08-15 | A kind of social platform user classification method based on the analysis of individual affective behavior |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610668449.4A CN106295702B (en) | 2016-08-15 | 2016-08-15 | A kind of social platform user classification method based on the analysis of individual affective behavior |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106295702A CN106295702A (en) | 2017-01-04 |
| CN106295702B true CN106295702B (en) | 2019-10-25 |
Family
ID=57670975
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610668449.4A Active CN106295702B (en) | 2016-08-15 | 2016-08-15 | A kind of social platform user classification method based on the analysis of individual affective behavior |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106295702B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107369099B (en) * | 2017-06-28 | 2021-01-22 | 江苏云机汇软件科技有限公司 | User behavior analysis system facing social network |
| CN107563429B (en) * | 2017-07-27 | 2020-11-10 | 国家计算机网络与信息安全管理中心 | Method and device for classifying network user groups |
| CN107608792B (en) * | 2017-09-12 | 2020-09-01 | 中国联合网络通信集团有限公司 | Resource scheduling method and device |
| CN107590742B (en) * | 2017-10-16 | 2021-06-22 | 东北大学 | Behavior-based social network user attribute value inversion method |
| CN108268624B (en) * | 2018-01-10 | 2020-04-24 | 华控清交信息科技(北京)有限公司 | User data visualization method and system |
| CN109271634B (en) * | 2018-09-17 | 2022-07-01 | 重庆理工大学 | Microblog text emotion polarity analysis method based on user emotion tendency perception |
| CN111565322B (en) * | 2020-05-14 | 2022-03-04 | 北京奇艺世纪科技有限公司 | User emotional tendency information obtaining method and device and electronic equipment |
| CN113158082B (en) * | 2021-05-13 | 2023-01-17 | 和鸿广科技(上海)有限公司 | Artificial intelligence-based media content reality degree analysis method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103678613A (en) * | 2013-12-17 | 2014-03-26 | 北京启明星辰信息安全技术有限公司 | Method and device for calculating influence data |
| CN105320960A (en) * | 2015-10-14 | 2016-02-10 | 北京航空航天大学 | Voting based classification method for cross-language subjective and objective sentiments |
| CN105631748A (en) * | 2015-12-21 | 2016-06-01 | 西北工业大学 | Parallel label propagation-based heterogeneous network community discovery method |
| CN105654115A (en) * | 2015-12-28 | 2016-06-08 | 西北工业大学 | Density adaptive clustering method orienting behavior identification |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130073480A1 (en) * | 2011-03-22 | 2013-03-21 | Lionel Alberti | Real time cross correlation of intensity and sentiment from social media messages |
-
2016
- 2016-08-15 CN CN201610668449.4A patent/CN106295702B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103678613A (en) * | 2013-12-17 | 2014-03-26 | 北京启明星辰信息安全技术有限公司 | Method and device for calculating influence data |
| CN105320960A (en) * | 2015-10-14 | 2016-02-10 | 北京航空航天大学 | Voting based classification method for cross-language subjective and objective sentiments |
| CN105631748A (en) * | 2015-12-21 | 2016-06-01 | 西北工业大学 | Parallel label propagation-based heterogeneous network community discovery method |
| CN105654115A (en) * | 2015-12-28 | 2016-06-08 | 西北工业大学 | Density adaptive clustering method orienting behavior identification |
Non-Patent Citations (5)
| Title |
|---|
| Cross-Domain and Cross-Category Emotion Tagging for Comments of Online News;Ying Zhang等;《proceedings SIGIR’14 proceedings of the 37TH international ACM SIGIR conference on research&development in information retrieve》;20140711;第627-636页 * |
| Discovering Information Propagation Patterns in Microblogging Services;ZHIWEN YU等;《ACM Transactions on Knowledge Discovery from Data》;20150731;第10卷(第1期);摘要、第3节、第6.1节、第7.2节，图1 * |
| Featuring, Detecting, and Visualizing Human Sentiment in Chinese Micro-Blog;ZHIWEN YU等;《ACM Transactions on Knowledge Discovery from Data》;20160530;第10卷(第4期);第48:1至48:23页 * |
| Lexicon-based Sentiment Analysis on Topical Chinese Microblog Messages;CUI Anqi等;《semantic web and science》;20130502;第1节、第2节，图1 * |
| Sentiment Detection and Visualization of Chinese Micro-blog;Zhitao Wang等;《2014 international conference on data science and advanced analytics(DSAA)》;20141101;第1-7页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106295702A (en) | 2017-01-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106295702B (en) | A kind of social platform user classification method based on the analysis of individual affective behavior | |
| Emon et al. | A deep learning approach to detect abusive bengali text | |
| CN105740228B (en) | A kind of internet public feelings analysis method and system | |
| CN104268197B (en) | A kind of industry comment data fine granularity sentiment analysis method | |
| CN103559233B (en) | Network neologisms abstracting method and microblog emotional analysis method and system in microblogging | |
| CN104615608B (en) | A kind of data mining processing system and method | |
| CN103678670B (en) | Micro-blog hot word and hot topic mining system and method | |
| CN105893582B (en) | A kind of social network user mood method of discrimination | |
| CN107688576B (en) | Construction and tendency classification method of CNN-SVM model | |
| CN112199608A (en) | Social media rumor detection method based on network information propagation graph modeling | |
| CN106202053B (en) | A kind of microblogging theme sentiment analysis method of social networks driving | |
| CN106202584A (en) | A kind of microblog emotional based on standard dictionary and semantic rule analyzes method | |
| CN109214454B (en) | Microblog-oriented emotion community classification method | |
| CN105354305A (en) | Online-rumor identification method and apparatus | |
| Celli et al. | Relationships between personality and interactions in facebook | |
| CN107305545A (en) | A kind of recognition methods of the network opinion leader based on text tendency analysis | |
| CN109726745A (en) | A kind of sensibility classification method based on target incorporating description knowledge | |
| CN110750648A (en) | Text emotion classification method based on deep learning and feature fusion | |
| CN110321918A (en) | The method of public opinion robot system sentiment analysis and image labeling based on microblogging | |
| CN112699240A (en) | Intelligent dynamic mining and classifying method for Chinese emotional characteristic words | |
| CN105956158B (en) | The method that network neologisms based on massive micro-blog text and user information automatically extract | |
| CN107392392A (en) | Microblogging forwarding Forecasting Methodology based on deep learning | |
| CN105869058B (en) | A kind of method that multilayer latent variable model user portrait extracts | |
| Wang et al. | A multidimensional nonnegative matrix factorization model for retweeting behavior prediction | |
| Kanev et al. | Sentiment analysis of multilingual texts using machine learning methods |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |