Managing Online Business Communities

Download Managing Online Business Communities

Post on 26-Jan-2015




0 download

Embed Size (px)


Keynote talk at Int. Conf. on Enterprise Information Systems, ICEIS-2012, Wroclaw, Poland


<ul><li> 1. Managing Online Business Communities The ROBUST Project Steffen Staab, Thomas Gottron&amp; ROBUST Project Team EC Project 257859</li></ul> <p> 2. Business Communities Information ecosystems Employees Business Partners, Customers General Public Valuable asset RisksOpportunities 3. Use Cases Lotus Connections SAP Community Network (SCN) MeaningMine Communities Communities Communities Employees Customers Social media Working groups Partners News Interest Groups Suppliers Web fora Projects Developers Public communities Business valueBusiness valueBusiness value Task relevant information Products support Topics Collaboration Services Opinions Innovation Find business partners Service for partners VolumeVolumeVolume 4,000 posts/day 6,000 posts/day 1,400,000 posts/day 386,000 employees 1,700,000 subscribers 708,000 web sources 1.5GB content/day 16GB log/day 45GB content/dayEmployees Business Partners Public Domain Intranet ExtranetInternet 4. High Level ArchitectureRisk Dashboard Community Community Risk Communication Bus SimulationManagement Cloud Based DataUser Roles Managementand Models Community Analysis Community Data 5. Requirements Community Risk Management Risk Formalization Community Analysis Risk Detection Risk Forecasting Feature Selection Risk Management Structural Analysis Risk Visualization Behavior Analysis Content Analysis User Roles and Models Cross Community User Needs Community Simulation Motivation Roles Community Models Groups Policy Models Community Value Influence SimulationCloud Based Data Management Prediction Operations on Data Scalability Real Time Parallel Execution Stream Based 6. The ROBUST platformRisk Dashboard and Visualizations 7. A new Role: the Community Manager Definition of risks/opportunities Monitoring Interaction with community Reaction and countermeasures Need for a control center 8. Risk matrix visualizationRisks ordered by probability Negative impacts Positive impacts A single risk can have multiple impacts 9. Community Health: Graphic Equalizer 10. Community Analysis,User Roles &amp; Risk ManagementDetermine Risk:Users Leaving the Community 11. Churn Users churn = Users leave the community, become inactive Questions: Why do users leave the community? Who is leaving the community? Impact? Tasks: Detect Churn Predict Churn Evaluate Churn Angeletou et al. ISWC 2011 12. What is Churn? Activity? No churnNo churn Activity burstsHoliday No churn Churn! Inconsistent Cease to be active 13. Why Churn? 14. Network Effects Churn: churning users influence other users they communicate with First Churner FirstinfluencedChurner Temporal ordering: Churners that churned subsequent to each other Frequently, after a slow start,Cascaderesulting in a cascade of churning Churn of a single userInfluences his neighboursKarnstedt et al. WebSci 2011 15. Role Analysis: who churns? A role represents the standing, or part, that a user has within a given community RoleDimensionReciprocity Initialisation Persistence Popularity Elitist high: Threadslow low: Neighbours Joining lowhigh Conversationalist Popular Initiator high high Supporter mediumlowmedium Taciturnlowlow Ignoredlow: No replies 16. Role Analysis: ApproachReciprocity, initialisation, Feature levels change with thepersistence, popularitydynamics of the communityBehaviourOntology Run rules over each users features Based on related work, we associate and derive the community role roles with a collection of feature-to-level composition mappings e.g. in-degree -&gt; high, out-degree -&gt; highAngeletou et al. ISWC 2011 17. Churn Analysis on Roles Role composition in community Development towards an unhealthy role composition! 18. Community &amp; Content Analysis Opportunity Detection:Discovering Interesting Contents 19. IBM Connections Analogies to Twitter! 20. Retweets RT @janedoe: MyFollowerdear @johndoehad troubles to wake up this@janedoe#morningMy dear @johndoe hadtroubles to wakeup this #morning 21. Retweets Retweet indicates quality of interest for others Idea: Learn to predict retweets! Likelihood of retweet as metric for Interestingness Naveed et al. WebSci 2011 22. Retweets: Prediction modelAim: Prediction of probability of retweetLogistic regression: 1f (z) =-z1+e z = w0 + w1x1 ++w2 x2 ++ wn xnModel parameters wi learned on training dataDataset Users TweetsRetweets Choudhury 118,506 9,998,7567.89% Choudhury (extended)277,66629,000,0008.64% Petrovic 4,050,944 21,477,4848.46% 23. Logistic Regression: WeightsFeature Dimensions WeightConstant(intercept)-5.45Direct message-147.89Username 146.82Message featureHashtag 42.27URL249.09Valence-26.88Sentiment Arousal 33.97Dominance 19.56Positive -21.8EmoticonsNegative9.94Positive13.66ExclamationNegative8.72!-16.85Punctuation? 23.67Terms Odds19.79 24. Logistic Regression: Topic Weights TopicWeightsocial media market post site web tool traffic network27.54follow thank twitter welcome hello check nice cool people 16.08credit money market business rate economy home15.25christmas shop tree xmas present today wrap finish2.87home work hour long wait airport week flight head -14.43twitter update facebook account page set squidoo check-14.43cold snow warm today degree weather winter morning-26.56night sleep work morning time bed feel tired home -75.19 25. Re-Ranking using Interestingness Top-k interesting tweets for beerRang Username Tweet 1 BeeracrossTX UK beer mag declares "the end of beer writing." @StanHieronymus says not so in the US. #beer 2narmmusicbeer summit @bspward @jhinderaker no one had billy beer? heehee #narm - beer summit @bspward @jhinde 3beeriety Go green and turn those empty beer bottles into recycled beer glasses! | #beer #recycle (via: @td333) 4hblackmonGreat Divide beer dinner @ Porter Beer Bar on 8/19 - $45 for 3 courses + beer pairings. 5nycraftbeerInteresting Concept-Beer launches&amp;hopes 2help craft beer drinkers enjoy beer they want @their fave pubs. 6carichardson Beer Cheddar Soup: Dish number two in my famed beer dinner series is Beer Cheddar Soup. I hadnt had too.. 7BeerBrewingNew York City Beer Events - Beer Tasting - New York Beer Festivals - New York Craft Beer #beer 8delphiforums Love beer? Our member is trying to build up a new beer drinkers forum. Grab a #beer and join us: 9Jamie_Mason #Baltimore Beer Week continues w/ a beer brkfst, beer pioneers luncheon, drink &amp; donateevent, beer tastings &amp; more. 10 carichardson Seattle and Beer: I went to Seattle last weekend. It was my friends stag - he likes beer - we drank beer.. et al. CIKM 2011 26. LiveTweet Che Alhadi et al. TREC 2011, ECIR 2012 27. Compare Interestingness to Detect Trends Tomorrow I am going to play tennis play golf do some gardening spread wisdom go to a Justin Bieber concert win the lottery 28. Community SimulationRisk Mitigation: SimulatingEffects of Policy Changes 29. Policies Governance of Communities Def.: Steering and coordinating actions ofcommunity members Implementation: Direct intervention of community owner Functionality of the community platform Mitigation of risks: Change platform functionality Impact? Schwagereit WebSci 2011 30. Governance by Policy ChangeWhat if using Policy 2 ?Policy 1Community ManagerPredictionPolicy 2 31. Scenario: Policy Controlled Content Generation Users generate content Users search content Users consume content Users interact with content Rate content Reply to content Where can a (platform) policy have effects? 32. Policy impacts on user behaviour Policy User Need for Content Consumption Content ContentRating Replying Selection Assessment User Need for Content Creation ContentContent CreationEvaluation Policy 33. Attention Management Part of Community: forum activity Goal: Steer user activity to specific forums User model parameters: Activity rate for creating threads Activity rate for creating replies Preferences for activity in specific forums Community Model Varied restrictions for thread creation Observed Metrics Response time on threads 34. Simulation based on observed parametersOnline CommunityMetrics of CommunityCommunityCommunity ArtifactsActivity Community CommunityMembers Platform (Persons) (Software)Analysis ComparisonAbstraction User ModelCommunity+ Platform ParametersModelMetrics ofSimulated SimulatedSimulated Community CommunityCommunityArtifacts ActivitySimulation 35. Variable Parameters indicated by different policy Users search content Presentation Ranking threads in content views Recency Social Closeness Topical closeness Popularity Observation: Influence which questions are answered Users generate content Restrict number of questions asked per forum Users turn to other questions Observation: Response time in some fora reduced 36. Backend TechnologiesIndexing DistributedSemantic Graphs 37. Community Information Examples Male persons who have a public profile document Computing science papers authored by social scientists American actors who are also politicians and are married to a model. Maybe specific databases available: Person search engines Bibliographic databases Movie database 38. Linked Data Semantic Web Technology to 1. Provide structured data on the web 2. Link data across data sourcesThingThingThingThingThingThingThingThingThingThingtypedtypedtypedtyped linkslinkslinkslinksA BC D E 39. The LOD Cloud 40. Querying linked data SELECT ?x WHERE {?x rdfs:type foaf:Person .?x rdfs:type pim:Male .?x foaf:maker ?y .?y rdfs:type foaf:PersonalProfileDocument . } 41. Querying linked data using an index SELECT ?x WHERE {?x rdfs:type foaf:Person .?x rdfs:type pim:Male .?x foaf:maker ?y .?y rdfs:type foaf:PersonalProfileDocument . } 42. Schema Index OverviewKonrath et al. BTC 2011, JWS 2012 43. How it works ...SELECT ?xFROM WHERE { ?x rdfs:type foaf:Person . ?x rdfs:type pim:Male . ?x foaf:maker ?y . ?y rdfs:type foaf:PersonalProfileDocument .}pim: foaf: foaf:MalePerson PPDrdfs:typerdfs:type rdfs:type?x?yfoaf:maker 44. How it works ...pim:Malefoaf:Personfoaf:PPD TC_18 TC_765_76maker EQC_5foaf:makerDS 1 DS 2 45. Building the Index from a Stream Stream of data (coming from a LD crawler) D16, D15, D14, D13, D12, D11, D10, D9, D8, D7, D6, D5, D4, D3, D2, D1 FiFo1 C346 C234 2 C2 2 13 C1 5 46. Does it work good?Comparison of stream based vs. Gold standard Schema on 11 M triple data set 47. Managing Business Communities Lessons learned 48. Conclusions Business communities vary with regard to Interaction Interests Type of conversation Novel analysis techniques needed: Integration of different data sources Simulation of policy changes Value of users Concrete needs confirmed by project external companies looking for such technology 49. References S. Angeletou, M. Rowe, H. Alani. Modelling and analysis of user behaviour in online communities.In Proceedings of the 10th international conference on The semantic web - Volume Part I, A. Che Alhadi, T. Gottron, J. Kunegis, N. Naveed. Livetweet: Microblog retrieval based oninterestingness. In TREC11: Proceedings of the Text Retrieval Conference 2011, 2011. A. Che Alhadi, T. Gottron, J. Kunegis, N. Naveed. Livetweet: Monitoring and predicting interestingmicroblog posts. In ECIR12: Proceedings of the 34th European Conference on Information Retrieval,2012. M. Karnstedt, M. Rowe, J. Chan, H. Alani, C. Hayes. The effect of user features on churn in socialnetworks. In WebSci 11: Proceedings of the 3rd International Conference on Web Science, 2011. M. Konrath, T. Gottron, A. Scherp. Schemex web-scale indexed schema extraction of linked opendata. In Semantic Web Challenge, Submission to the Billion Triple Track, 2011. M. Konrath, T. Gottron, S. Staab, A. Scherp. Schemexefficient construction of a data catalogue bystream-based indexing of linked data. Journal of Web Semantics, 2012. to appear. N. Naveed, T. Gottron, J. Kunegis, A. Che Alhadi. Bad news travel fast: A content-based analysis ofinterestingness on twitter. In WebSci 11: Proceedings of the 3rd International Conference on WebScience, 2011. N. Naveed, T. Gottron, J. Kunegis, A. Che Alhadi. Searching microblogs: Coping with sparsity anddocument quality. In CIKM11: Proceedings of 20th ACM Conference on Information and KnowledgeManagement, 2011. F. Schwagereit, A. Scherp, S. Staab. Survey on governance of user-generated content in webcommunities. In WebSci 11: Proceedings of the 3rd International Conference on Web Science, 2011. 50. Thank You! EC Project 257859 </p>