Collective Spammer Detection in Evolving Multi-Relational Social NetworksShobeir

Collective Spammer Detection in Evolving Multi-Relational Social NetworksShobeir www.phwiki.com

Collective Spammer Detection in Evolving Multi-Relational Social NetworksShobeir

Halverstadt, Lisa, Police Reporter has reference to this Academic Journal, PHwiki organized this Journal Collective Spammer Detection in Evolving Multi-Relational Social NetworksShobeir Fakhraei (University of Maryl in addition to ) James Foulds (University of Cali as long as nia, Santa Cruz) Madhusudana Shashanka (if(we) Inc., Currently Niara Inc.) Lise Getoor (University of Cali as long as nia, Santa Cruz)Spam in Social NetworksRecent study by Nexgate in 2013:Spam grew by more than 300% in half a year2Spam in Social NetworksRecent study by Nexgate in 2013:Spam grew by more than 300% in half a year1 in 200 social messages are spam3

Logan College of Chiropractic US www.phwiki.com

This Particular University is Related to this Particular Journal

Spam in Social NetworksRecent study by Nexgate in 2013:Spam grew by more than 300% in half a year1 in 200 social messages are spam5% of all social apps are spammy 4Spam in Social NetworksWhat’s different about social networksSpammers have more ways to interact with users 5Spam in Social NetworksWhat’s different about social networksSpammers have more ways to interact with users Messages, comments on photos, winks, 6

Spam in Social NetworksWhat’s different about social networksSpammers have more ways to interact with users Messages, comments on photos, winks, They can split spam across multiple messages7Spam in Social NetworksWhat’s different about social networksSpammers have more ways to interact with users Messages, comments on photos, winks, They can split spam across multiple messagesMore available info about users on their profiles!8Spammers are getting smarter!9Want some replica luxury watches Click here: http://SpammyLink.comTraditional Spam:

Spammers are getting smarter!10Want some replica luxury watches Click here: http://SpammyLink.comTraditional Spam: [Report Spam]Spammers are getting smarter!11Want some replica luxury watches Click here: http://SpammyLink.comTraditional Spam:(Intelligent) Social Spam:Hey Shobeir!Nice profile photo. I live in Bay Area too. Wanna chat [Report Spam]Spammers are getting smarter!12Want some replica luxury watches Click here: http://SpammyLink.comTraditional Spam:(Intelligent) Social Spam:Hey Shobeir!Nice profile photo. I live in Bay Area too. Wanna chatSure! 🙂 [Report Spam]

Spammers are getting smarter!13Want some replica luxury watches Click here: http://SpammyLink.comTraditional Spam:(Intelligent) Social Spam:Hey Shobeir!Nice profile photo. I live in Bay Area too. Wanna chatSure! 🙂 [Report Spam] I’m logging off here., too many people pinging me! I really like you, let’s chat more here:http://SpammyLink.comRealistic Looking ConversationTagged.comFounded in 2004, is a social networking site which connects people through social interactions in addition to gamesOver 300 million registered membersData sample as long as experiments (on a laptop):5.6 Million users (3.9% Labeled Spammers)912 Million Links14Social Networks: Multi-relational in addition to Time-Evolving15

Social Networks: Multi-relational in addition to Time-Evolving16Legitimate usersSocial Networks: Multi-relational in addition to Time-Evolving17Legitimate usersSpammersSocial Networks: Multi-relational in addition to Time-EvolvingLink = Action at time tActions = Profile view, message, poke, report abuse, etc18Legitimate usersSpammers

Social Networks: Multi-relational in addition to Time-EvolvingLink = Action at time tActions = Profile view, message, poke, report abuse, etc19Social Networks: Multi-relational in addition to Time-EvolvingLink = Action at time tActions = Profile view, message, poke, report abuse, etc20Profile viewSocial Networks: Multi-relational in addition to Time-EvolvingLink = Action at time tActions = Profile view, message, poke, report abuse, etc21Profile viewMessage

Social Networks: Multi-relational in addition to Time-EvolvingLink = Action at time tActions = Profile view, message, poke, report abuse, etc22Profile viewMessagePokeSocial Networks: Multi-relational in addition to Time-EvolvingLink = Action at time tActions = Profile view, message, poke, report abuse, etc23Profile viewMessagePokeReportspammerOur Approach24Predict spammers based on:Graph structureAction sequencesReporting behavior

Halverstadt, Lisa Arizona Republic - Mesa Bureau, The Police Reporter www.phwiki.com

Our Approach25Predict spammers based on:Graph structureAction sequencesReporting behavior Graph Structure Feature Extraction26Graphs as long as each relationGraph Structure Feature Extraction27Graphs as long as each relationFeatures

Graph Structure FeaturesExtract features as long as each relation graph es as long as each of 10 relPageRankDegree statisticsTotal degree In degreeOut degree k-CoreGraph coloringConnected componentsTriangle count28(8 features as long as each of 10 relations) Graph Structure FeaturesExtract features as long as each relation graph es as long as each of 10 relPageRankDegree statisticsTotal degree In degreeOut degree k-CoreGraph coloringConnected componentsTriangle count29(8 features as long as each of 10 relations) Graph Structure FeaturesExtract features as long as each relation graph es as long as each of 10 relPageRankDegree statisticsTotal degree In degreeOut degree k-CoreGraph coloringConnected componentsTriangle count30(8 features as long as each of 10 relations)

Acknowledgements Collaborators:If(we) Inc. (Formerly Tagged Inc.): Johann Schleier-Smith, Karl Dawson, Dai Li, Stuart Robinson, Vinit Garg, in addition to Simon HillDato (Formerly Graphlab): Danny Bickson, Brian Kent, Srikrishna Sridhar, Rajat Arya, Shawn Scully, in addition to Alice Zheng64Shobeir FakhraeiUniv. of Maryl in addition to Lise GetoorUniv. Cali as long as nia, Santa CruzMadhusudana Shashankaif(we) Inc., currently Niara Inc.Conclusion65Report SubgraphProbabilistic Soft LogicMultiple relations are more predictive than multiple featuresEven simple bigrams are highly predictive Jointly refining the credibility of the source is highly effective!AUPR:0.187 0.328 AUPR: 0.471AUPR:0.674 0.884Can classify 70% of the spammers that needed manual labeling with 90% accuracy AUPR: 0.779Code in addition to part of the data will be released soon: https://github.com/shobeir/fakhraei-kdd2015 Thank you!

Halverstadt, Lisa Police Reporter

Halverstadt, Lisa is from United States and they belong to Arizona Republic – Mesa Bureau, The and they are from  Mesa, United States got related to this Particular Journal. and Halverstadt, Lisa deal with the subjects like Local News; Police and Law Enforcement

Journal Ratings by Logan College of Chiropractic

This Particular Journal got reviewed and rated by Logan College of Chiropractic and short form of this particular Institution is US and gave this Journal an Excellent Rating.