Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
MAKE GROUP DETECTION MORE HUMAN Motahareh EslamiMehdiabadi Cutlure as Data Fall 2012 GROUPS Communities, clusters or modules Community structure: many relations within a group/ few relations between groups Independent Compartments Detecting groups (communities) Sociology Biology Computer Science Hard problem Not yet satisfactorily solved! 2 WHY DETECTING GROUPS? Many real networks have community structures. Families Friendship circles Villages and Towns Virtual groups on internet … Clustering web clients who are geographically near to each other Identifying clusters of customers with similar interests Ad-hoc networks Classification of vertices 3 THE CHALLENGE Several group detection algorithms No one cannot state which method (or subset of methods) is the most reliable one in applications. Testing and Evaluation Using simple benchmark graphs A LFR benchmark graph Debating over complexity and time Limited evaluation measures 4 A NEW APPROACH OF EVALUATION Asking people to evaluate! Facebook Network Use efficient and popular algorithms Grivan-Newman (GN) Markov Clustering (MCL) Clauset-Newman –More (CNM) 5 DATA Step 1:Interview Name the clusters Change the clustering as they want Tell us their idea! ……. Step 2:Online Application Join us soon…! 6 CDA 7 REFERENCES Fortunato, Santo. Community detection in graphs. Physics Reports 486.3 (2010): 75-174. Lancichinetti, Andrea, Santo Fortunato, and Filippo Radicchi. Benchmark graphs for testing community detection algorithms. Physical Review E 78.4 (2008): 046110. Han, Jiawei, and Micheline Kamber. Data mining: concepts and techniques. Morgan Kaufmann, 2006. … 8 QUESTIONS? 9