CN103902621B - Method and device for identifying network rumor - Google Patents

Method and device for identifying network rumor Download PDF

Info

Publication number
CN103902621B
CN103902621B CN201210586904.8A CN201210586904A CN103902621B CN 103902621 B CN103902621 B CN 103902621B CN 201210586904 A CN201210586904 A CN 201210586904A CN 103902621 B CN103902621 B CN 103902621B
Authority
CN
China
Prior art keywords
rumour
network
network information
user
database
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
Application number
CN201210586904.8A
Other languages
Chinese (zh)
Other versions
CN103902621A (en
Inventor
魏彦杰
张帆
张慧玲
彭丰斌
孟金涛
魏丹
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.)
Shenzhen Institute of Advanced Technology of CAS
Original Assignee
Shenzhen Institute of Advanced Technology of CAS
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 Shenzhen Institute of Advanced Technology of CAS filed Critical Shenzhen Institute of Advanced Technology of CAS
Priority to CN201210586904.8A priority Critical patent/CN103902621B/en
Publication of CN103902621A publication Critical patent/CN103902621A/en
Application granted granted Critical
Publication of CN103902621B publication Critical patent/CN103902621B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90Details of database functions independent of the retrieved data types
    • G06F16/95Retrieval from the web

Abstract

The invention belongs to the field of Internet communications and provides a method for identifying a network rumor. The method comprises the steps that network information in a database is analyzed, and features are extracted; a model is built through the machine learning method to generate a scoring function; the network information is identified by utilizing the scoring function. According to the method, the model can be amended at intervals, the dynamism of network communication is embodied; the scoring function generated through model building can quickly identify the network rumor by utilizing a machine, and therefore an important basis is provided for a quick response of network administrators.

Description

A kind of method and apparatus of identification network rumour
Technical field
The invention belongs to field of Internet communication, more particularly, to a kind of method and apparatus of identification network rumour.
Background technology
Developing rapidly with Internet technology, Facebook, Twitter, microblogging, Email, blog, youtube In succession occur Deng social network sites, network rumour also produces therewith and propagates, its harm that society is produced:Little to personal lifestyle and Privacy, the big stable and united and economic development arriving society.In such as 2012 summer, there is sanguinary incident in Assam of India Afterwards, network rumour leads to more than 30 ten thousand people to flee from residence;On 2 20th, 2010, Shanxi certain areas wanted the rumour of earthquake to pass through Network is propagated rapidly, causes ground millions of masses in Taiyuan etc. six to start in morning to go on street corner " hiding earthquake ", earthquake research in shanxi official website Once paralysed;After the 9.0 grades of earthquakes of in March, 2011 Japan, relevant salt can make China Partial ground with the rumour of prevention of nuclear Area starts mad panic buying salt, market order complete confusion.According to statistics only 2012 between March and April in the net that China is cleaned Network rumour just has more than 20 ten thousand a plurality of.Network rumour crosses the boundary of a country, and worldwide affects and harm the living a stable life of the people, society Meeting safety and economic development, administer network rumour and have become as a global difficult problem.
Network rumour can be understood as under this specific environment in network, and Web vector graphic entity is propagated in a specific way , things interested in netizen, event or problem, unverified illustrate or annotate (《Theory and Application》2004 6 phases of year, nest is roc, Huang Xianzhu).Also it is not directed to one kind fast and effectively side of identification of network rumour at present in world wide How method, identify that network rumour meaning is very great at short notice.The patent of Application No. 200810167018.5 describes A kind of network order regulating method, the method carries out specification mainly for the network behavior of the network user, based on the network user's A credit evaluation system is set up in behavior, is not directed to network rumour and proposes effectively prediction and authentication method.
Content of the invention
The embodiment of the present invention provides a kind of method and apparatus of identification network rumour it is intended to solve currently without for network Rumour proposes effectively prediction and authentication method it is impossible to utilize machine Rapid identification network rumour, thus fast for network manager Speed reaction provides foundation.
For this reason, embodiments providing following technical scheme:
A kind of method of identification network rumour, comprises the following steps:
The network information in database is analyzed and extracts feature;
With the modeling of machine learning method, generate scoring functions;
Using scoring functions, the network information is identified;
Wherein, described step A comprises the following steps:
a:Classification to rumour in database, each rumour is classified as a class therein;
b:Each class rumour and each network user in analytical database, extract the feature relevant with communication environments;
c:Each class rumour in analytical database, extracts the feature related to the network information itself;
d:Extract the non-rumour network information of equal number from database, repeat described step a to step c, right The described non-rumour network information is analyzed and extracts feature;
Described step c comprises the following steps:
c1:According to significance level difference, rumour is classified, by the spread speed of rumour in statistical data analysis storehouse, determine The upper limit threshold of unit interval propagation times and lower threshold, and whether exceeded according to propagation times within the unit interval for the rumour Described upper limit threshold, be less than described lower threshold, or exceed described lower threshold and be less than described upper limit threshold, realize from biography Broadcast classifying importance feature and the importance analysis to rumour of VELOCITY EXTRACTION rumour;
c2:According to the fuzzy word list extracted from social network database and analysis obtains, and in rumour, fuzzy word goes out Existing frequency, the ambiguity classification to rumour, realize the ambiguity analysis to rumour;
c3:Be sent to the network user by automated randomized for rumour, according to the analysis to feedback information, to rumour abnormality Classification, realizes the unusual degree analysis to rumour.
Compared with prior art, embodiments of the invention have the advantage that:
The present invention is analyzed to the network information in database and extracts feature by providing, and is built with machine learning method Mould, generates scoring functions, recycles scoring functions that the network information is identified, often after a while model can be made and repairing Just, and realize using machine Rapid identification network rumour, thus providing important foundation for network manager fast reaction.
Brief description
Fig. 1 is the method flow diagram of identification network rumour provided in an embodiment of the present invention;
Fig. 2 is the structure chart of the device of identification network rumour provided in an embodiment of the present invention.
Specific embodiment
In order that the objects, technical solutions and advantages of the present invention become more apparent, below in conjunction with drawings and Examples, right The present invention is further elaborated.It should be appreciated that described herein be only the present invention a part of embodiment, rather than Whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art are not before making creative work Put the every other embodiment being obtained, broadly fall into the scope of protection of the invention.
Specifically, the Common Parameters used in some this patented inventions are defined first:
◆M:The network information, can be rumour or real information.
◆S(M):The publisher of network information M.
◆T(M):The disseminator of network information M, have propagated this network information;Can be the use having believed this network information Family or the user not believing that this network information.
◆R(M):The audient of network information M is it is believed that the network user of this network information;Can be network information M simultaneously Disseminator.
◆U:The network user, can be personal or unit mechanism;The main body of network interdynamic, can be that information is sent out Cloth person, disseminator and audient.
◆I(M):The importance of network information M.
◆V(M):The ambiguity of network information M.
◆A(M):The abnormality of network information M.
◆p1,p2,p3…pi…:The parameter of scoring functions, each parameter pi is directed to of scoring functions.
In the present invention, the scoring functions for certain network information M have following form:
F (M)=p1*S+p2*T+p3*R+p4*I+p5*V+p6*A (E1)
The function of scoring functions (E1) here is made up of two parts.The propagation of first three items and the network information and communication environments Relevant, then three features with the network information itself are relevant.
Fig. 1 is the method flow diagram of identification network rumour provided in an embodiment of the present invention, for convenience of description, illustrate only The part related to the embodiment of the present invention.
As shown in figure 1, the method comprises the following steps:
Step 101, is analyzed to the network information in database and extracts feature.
Specifically, comprise the following steps:
Step one, the classification to rumour in database, each rumour is classified as a class therein.
Specific it is assumed that being divided into NM class rumour, such as can be divided into finance and economics, physical culture, amusement, automobile, politics, science and technology, military affairs, History, other etc., then each rumour is classified as a class therein.
Step 2, each class rumour and each network user in analytical database, extract the spy relevant with communication environments Levy.
Specifically, the communication environments of the network information comprise the publisher of the network information, disseminator, and audient.Because network Information (rumour) communication environments are extremely important, and only through wide-scale distribution, its impact and harm just can be huge.Disseminator and audient The propagation of the network information can be largely effected on to the judgement of this event.Such as:Earthquake research in shanxi office is as the network user (such as Microblog users) want the rumour of earthquake to refute a rumour 2 months Shanxi in 2010 certain areas, prevent being propagated further of rumour.
Specifically, the feature relevant with communication environments, in the parameter that the present invention uses, is defined as:
◆S(M):The publisher of network information M.
◆T(M):The disseminator of network information M, have propagated this network information;Can be the use having believed this network information Family or the user not believing that this network information.
◆R(M):The audient of network information M is it is believed that the network user of this network information;Can be network information M simultaneously Disseminator.
Preferably, for each class rumour Mi in database, and each network user Ui makees following analysis, and extracts The feature relevant with communication environments.
First, whether rumour Mi was issued according to user, user's mark was rumour publisher/non-published person, thus right The publisher of the network information is analyzed.
Preferably, the publisher S of the network information is analyzed.Because user Ui is probably the publisher of network rumour, It is likely to not be therefore, whether to issue rumour Mi according to user Ui, user Ui is labeled as:Mi rumour publisher/non-published Person.Therefore for the user Ui in database, corresponding scoring functions:
F (M)=p1*S+p2*T+p3*R+p4*I+p5*V+p6*A (E1)
In E1, S item just has two, and p1 parameter also just has two.S item ginseng corresponding to all users and all rumour types The sum of number is 2*N*L.
Secondly, whether rumour Mi is propagated through according to user, user's mark is gossip propagation person/non-propagating person, thus right The disseminator of the network information is analyzed.
Preferably, the disseminator T of the network information is analyzed.Because user Ui is probably the disseminator of network rumour, It is likely to not be therefore, whether to be propagated through rumour Mi mark Ui according to user Ui:Mi gossip propagation person/non-propagating person;Therefore right User Ui in database, corresponding scoring functions:
F (M)=p1*S+p2*T+p3*R+p4*I+p5*V+p6*A (E1)
In E1, T item just has two, and p2 parameter also just has two.T item ginseng corresponding to all users and all rumour types The sum of number is 2*N*L.
Then, whether rumour Mi is believed according to user, by user's mark rumour audient/non-audient, thus to the network information Audient be analyzed.
The audient R of the network information is analyzed.Because user Ui may believe the network rumour receiving it is also possible to not Believe the network rumour receiving, therefore whether believe that rumour Mi mark user Ui is according to user Ui:Mi rumour audient/non-is subject to Many;Therefore for the user Ui in database, corresponding scoring functions:
F (M)=p1*S+p2*T+p3*R+p4*I+p5*V+p6*A (E1)
In E1, R item just has two, and p3 parameter also just has two.T item ginseng corresponding to all users and all rumour types The sum of number is 2*N*L.
Step 3, each class rumour in analytical database, extracts the feature related to the network information itself.
Preferably, comprise the following steps:
First, according to significance level difference, rumour is classified, by the spread speed of rumour in statistical data analysis storehouse, really Determine upper limit threshold and the lower threshold of unit interval propagation times, and whether surpassed according to propagation times within the unit interval for the rumour Cross described upper limit threshold, be less than described lower threshold, or exceed described lower threshold and be less than described upper limit threshold, realize from Spread speed extracts the classifying importance feature of rumour and the importance analysis to rumour.
Preferably, importance I of rumour is analyzed.Rumour can be divided into following a few class in general:Complaint ballad Speech, aggressive rumour, publicity property rumour, making profit property rumour, the property misread rumour, the important and harmfulness of different rumours is also different, According to its significance level difference can by non-for rumour be three classes:Extremely important, typically important, inessential.For hierarchy of operation, Characteristic of division can be extracted from spread speed, when propagation times in time t for the rumour Mi are more than Q1, this rumour Mi is non- Often important;When propagation times are less than Q1 more than Q2, this rumour Mi is typically important;When propagation times are less than Q2, this ballad Speech Mi is typically important.Wherein Q1>Q2, determines the value of Q1 and Q2 by the spread speed of rumour in statistical data analysis storehouse. Sum corresponding to the I item parameter of all rumour types is 3*L.
Secondly, according to the fuzzy word list extracted from social network database and analysis obtains, and fuzzy word in rumour The frequency occurring, the ambiguity classification to rumour, realize the ambiguity analysis to rumour.
Preferably, ambiguity V of rumour is analyzed.It is true to confirm that the low rumour of ambiguity is easy to the network user Vacation, therefore spread speed are slow;And the spread speed of the high rumour of ambiguity then will faster, its harm is also bigger.Can be according to ballad The ambiguity of rumour is divided three classes by the frequency calling the turn fuzzy word appearance:Ambiguity is high, and in ambiguity, ambiguity is low.Fuzzy word List need to extract from social network database and analyze, two frequency threshold V1 and V2 guiding point is arranged based on statistical analysis Class.Sum corresponding to the V item parameter of all rumour types is 3*L (design parameter is undetermined).According to from social network database The middle list extracted and analyze the fuzzy word obtaining, and the frequency that in rumour, fuzzy word occurs, the ambiguity of rumour is divided into three Class:Ambiguity is high, and in ambiguity, ambiguity is low.
Finally, be sent to the network user by automated randomized for rumour, according to the analysis to feedback information, to rumour abnormality Property classification, realize the unusual degree to rumour and analyze.
Preferably, the unusual degree A of rumour is analyzed.Rumour abnormality degree is higher, and its harm is bigger, and spread speed is got over Hurry up.The abnormality of rumour is divided three classes:Unusual degree is high, and in unusual degree, unusual degree is low.Concrete grammar is design one network system It is sent to some network users by automated randomized for rumour, rumour is classified by foundation to the analysis of feedback information.Corresponding to institute The sum having the A item parameter of rumour type is 3*L.
Step 4, extracts the non-rumour network information of equal number from database.
Specifically, find non-rumour network information L item, the step repeating 1-3, to L item Analysis of Network Information and extract spy Levy.
Step 102, with the modeling of machine learning method, generates scoring functions..
Specifically, including:
Step one, prepares sample characteristics, the rumour sample of acquisition and non-rumour sample and sample characteristics is changed into corresponding machine The form of device study classification method.
Step 2, uses machine learning classification method, and the rumour sample to described acquisition and non-rumour sample carry out many re-examinations Card training modeling, thus obtain the parameter of scoring functions model.
Specifically, described use machine learning classification method, including SVMs, one of neutral net or many Kind.
Step 103, is identified to the network information using scoring functions.
Specifically, comprise the following steps:
To any one new network information in database, extract the feature relevant with communication environments, and with the network information originally The related feature of body;
Specifically, following 6 features in the new network information are extracted:
S:The publisher of the network information.
T:The disseminator of the network information, have propagated this network information.
R:The audient of the network information.
I:The importance of the network information.
V:The ambiguity of the network information.
A:The abnormality of network information M.
Given a mark using scoring functions, specifically, scoring functions are:
F (M)=p1*S+p2*T+p3*R+p4*I+p5*V+p6*A (E1)
Due to modeling through machine learning, parameter p1 of function (E1), p2, p3 ... pi ..., calculate.Therefore, The fraction of scoring functions can be obtained.Thus identifying whether the described network information is rumour according to fraction.
Specifically, when described fraction is higher than a high preset value, then identify that this network information is rumour, less than one During low preset value, then identify that this network information is not network rumour, when fraction is between described high preset value and low default When between value, then defining this network information there is a strong possibility is network rumour, needs more information to verify further.
Preferably, the information that network is propagated is given a mark, when fraction is higher than certain threshold value F1, then identify this network Information is rumour;During less than certain threshold value F2, then identify that this network information is not network rumour;When fraction is between F1 and F2 When, then defining this network information there is a strong possibility is network rumour, needs more information to verify further.
Based on identical design, the embodiment of the present invention also provides a kind of device of identification network rumour, as shown in Fig. 2 should Device includes:
Database 201, for storing the network information.
Characteristic extracting module 202, for being analyzed to the network information in database and extracting feature.
MBM 203, for the modeling of machine learning method, generating scoring functions;
Identification module 204, for being predicted to the network information using scoring functions.
The embodiment of the present invention is analyzed to the network information in database and extracts feature by providing, and uses machine learning Method models, and generates scoring functions, recycles scoring functions that the network information is identified, can often after a while model be done Go out to revise, and realize using machine Rapid identification network rumour, thus providing important foundation for network manager fast reaction.
It will be appreciated by those skilled in the art that module in device in embodiment can be carried out point according to embodiment description It is distributed in the device of embodiment and be disposed other than in one or more devices of the present embodiment it is also possible to carry out respective change.On The module stating embodiment can merge into a module it is also possible to be further split into multiple submodule.
Through the above description of the embodiments, those skilled in the art can be understood that the present invention can be by Software adds the mode of necessary general hardware platform to realize naturally it is also possible to pass through hardware, but the former is more in many cases Good embodiment.Based on such understanding, technical scheme substantially contributes to prior art in other words Partly can be embodied in the form of software product, this computer software product is stored in a storage medium, if including Dry instruction is with so that a station terminal equipment (can be mobile phone, personal computer, server, or network equipment etc.) executes basis Invent the method described in each embodiment.
The above is only the preferred embodiment of the present invention it is noted that ordinary skill people for the art For member, under the premise without departing from the principles of the invention, some improvements and modifications can also be made, these improvements and modifications also should Depending on protection scope of the present invention.

Claims (5)

1. a kind of method of identification network rumour is it is characterised in that comprise the following steps:
A:The network information in database is analyzed and extracts feature;
B:With the modeling of machine learning method, generate scoring functions;
C:Using scoring functions, the network information is identified;
Wherein, described step A comprises the following steps:
a:Classification to rumour in database, each rumour is classified as a class therein;
b:Each class rumour and each network user in analytical database, extract the feature relevant with communication environments;
c:Each class rumour in analytical database, extracts the feature related to the network information itself;
d:Extract the non-rumour network information of equal number from database, repeat described step a to step c, to described The non-rumour network information is analyzed and extracts feature;
Described step c comprises the following steps:
c1:According to significance level difference, rumour is classified, by the spread speed of rumour in statistical data analysis storehouse, determine unit The upper limit threshold of time propagation times and lower threshold, and according to whether propagation times within the unit interval for the rumour exceed Upper limit threshold, be less than described lower threshold, or exceed described lower threshold and be less than described upper limit threshold, realize from propagate speed Degree extracts the classifying importance feature of rumour and the importance analysis to rumour;
c2:According to the fuzzy word list extracted from social network database and analysis obtains, and fuzzy word appearance in rumour Frequency, the ambiguity classification to rumour, realize the ambiguity analysis to rumour;
c3:Be sent to the network user by automated randomized for rumour, according to the analysis to feedback information, to rumour abnormality divide Class, realizes the unusual degree analysis to rumour.
2. the method for identification network rumour as claimed in claim 1 is it is characterised in that described step b comprises the following steps:
b1:Whether rumour was issued according to user, user's mark was rumour publisher/non-published person, thus to the network information Publisher be analyzed;
b2:Whether rumour is propagated through according to user, user's mark is gossip propagation person/non-propagating person, thus to the network information Disseminator be analyzed;
b3:Whether rumour is believed according to user, by user's mark rumour audient/non-audient, thus entering to the audient of the network information Row analysis.
3. the method for identification network rumour as claimed in claim 1 is it is characterised in that described step B comprises the following steps:
e:Prepare sample characteristics, the rumour sample of acquisition and non-rumour sample and sample characteristics are changed into corresponding machine learning classification The form of method;
f:Use machine learning classification method, the rumour sample to described acquisition and non-rumour sample carry out multiple-authentication training and build Mould, thus obtain the parameter of scoring functions model.
4. the method for identification network rumour as claimed in claim 3 is it is characterised in that described step C comprises the following steps:
g:To any one new network information in database, extract feature;
h:Given a mark using scoring functions, identified whether the described network information is rumour according to fraction.
5. the method for identification network rumour as claimed in claim 4 is it is characterised in that described step h comprises the following steps:
When described fraction is higher than a high preset value, then identify that this network information is rumour, the preset value low less than When, then identify that this network information is not network rumour, when fraction is between when between described high preset value and low preset value, then Defining this network information there is a strong possibility is network rumour, needs more information to verify further.
CN201210586904.8A 2012-12-28 2012-12-28 Method and device for identifying network rumor Active CN103902621B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201210586904.8A CN103902621B (en) 2012-12-28 2012-12-28 Method and device for identifying network rumor

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201210586904.8A CN103902621B (en) 2012-12-28 2012-12-28 Method and device for identifying network rumor

Publications (2)

Publication Number Publication Date
CN103902621A CN103902621A (en) 2014-07-02
CN103902621B true CN103902621B (en) 2017-02-08

Family

ID=50993949

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201210586904.8A Active CN103902621B (en) 2012-12-28 2012-12-28 Method and device for identifying network rumor

Country Status (1)

Country Link
CN (1) CN103902621B (en)

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104301330B (en) * 2014-10-29 2017-09-15 云南大学 The Honey net detection method measured with member's cohesion is monitored based on abnormal behaviour
CN105045857A (en) * 2015-07-09 2015-11-11 中国科学院计算技术研究所 Social network rumor recognition method and system
CN105787101B (en) * 2016-03-18 2019-06-07 联想(北京)有限公司 A kind of information processing method and electronic equipment
CN107644029A (en) * 2016-07-20 2018-01-30 平安科技(深圳)有限公司 Information query method and information query device
CN107797998B (en) * 2016-08-29 2021-05-07 腾讯科技(深圳)有限公司 Rumor-containing user generated content identification method and device
CN106570162B (en) * 2016-11-04 2020-07-28 北京百度网讯科技有限公司 Artificial intelligence-based rumor recognition method and device
CN106599286A (en) * 2016-12-23 2017-04-26 北京奇虎科技有限公司 Information monitoring rumor refuting realization method and apparatus, and mobile terminal
CN109327488B (en) * 2017-07-31 2021-08-13 腾讯科技(深圳)有限公司 Network information pushing method and device and server
CN108830630B (en) * 2018-04-09 2020-04-10 平安科技(深圳)有限公司 False message identification method and equipment
CN111079444B (en) * 2019-12-25 2020-09-29 北京中科研究院 Network rumor detection method based on multi-modal relationship
CN112417258A (en) * 2020-12-02 2021-02-26 深圳市罗湖医院集团 Crushing method, platform and terminal for rumor information in health knowledge search engine
CN113434684B (en) * 2021-07-01 2022-03-08 北京中科研究院 Rumor detection method, system, equipment and storage medium for self-supervision learning

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6687703B2 (en) * 1997-12-26 2004-02-03 Matsushita Electric Industrial Co., Ltd. Information filtering system and information filtering method
CN102609475A (en) * 2012-01-19 2012-07-25 浙江省公众信息产业有限公司 Method for monitoring content of microblog and monitoring system
CN102779174A (en) * 2012-06-26 2012-11-14 北京奇虎科技有限公司 Public opinion information display system and method

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6687703B2 (en) * 1997-12-26 2004-02-03 Matsushita Electric Industrial Co., Ltd. Information filtering system and information filtering method
CN102609475A (en) * 2012-01-19 2012-07-25 浙江省公众信息产业有限公司 Method for monitoring content of microblog and monitoring system
CN102779174A (en) * 2012-06-26 2012-11-14 北京奇虎科技有限公司 Public opinion information display system and method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Automatic Detection of Rumor on Sina Weibo;Fan Yang等;《MDS’12 Proceedings of the ACM SIGKDD Workshop on Mining Data Semantics》;20120812;摘要,第4部分,第5部分,表1 *
Rumor has it: Identifying Misinformation in Microblogs;Vehed Qazvinian 等;《EMNPL"11 Proceedings of the Conference on Empirical Methods in Natural Language Processing》;20110727;1589-1599 *

Also Published As

Publication number Publication date
CN103902621A (en) 2014-07-02

Similar Documents

Publication Publication Date Title
CN103902621B (en) Method and device for identifying network rumor
Liu et al. Weighted graph clustering for community detection of large social networks
CN102970289B (en) The identity identifying method of sing on web user behavior pattern
CN109479061A (en) Compliance violates detection
CN102819701B (en) Method and device for prompting password security of form filling component
CN105229485A (en) Multifactor location verification
CN104615608A (en) Data mining processing system and method
CN106682152A (en) Recommendation method for personalized information
CN105302810A (en) Information search method and apparatus
CN106294314A (en) Topics Crawling method and device
CN103577404A (en) Microblog-oriented discovery method for new emergencies
CN104408033A (en) Text message extracting method and system
CN110275965A (en) Pseudo event detection method, electronic device and computer readable storage medium
CN103577531A (en) Message subscription system and method based on message aggregate characteristics
CN112633973A (en) Commodity recommendation method and related equipment thereof
CN104778283A (en) User occupation classification method and system based on microblog
WO2014113405A2 (en) Systems and methods for processing and displaying user-generated content
CN110135978A (en) User's financial risks appraisal procedure, device, electronic equipment and readable medium
CN109634570A (en) Front and back end integrated development method, device, equipment and computer readable storage medium
CN105989261A (en) Method for securing electronic device and computer system
CN110516062A (en) A kind of search processing method and device of document
Saquib et al. Understanding dynamics of trending topics in Twitter
CN105426392A (en) Collaborative filtering recommendation method and system
CN102855130A (en) Client software classification method based on human needs
CN104809236B (en) A kind of age of user sorting technique and system based on microblogging

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant