View - Ecole Centrale Paris

DB2SNA: Extraction and Aggregation of Social Networks from DB 3theiraccountantcolleagues,withthesalesmen,withtheworkersand,finallywiththeir manager. And all these people play, of course, different roles in the businessof their enterprise. That is why we propose in section 3 to use heterogeneousgraphs to model and extract such complex social networks, by working directlyon the resulting graph database. The extraction of the social network is madeusing graph transformations. In a last step, we propose a visualization processbecause the ”raw data” of a network can have, in general, a huge size whichmakes its difficult to analyze and visualize. In order to ease the latter tasks wepropose to use an aggregation process in section 4. After a study of the existinggraph aggregation methods, we propose to use an existing technique which takesinto account the heterogeneity of our networks. A global view of our all-in-onesolution is given in Figure 1. Finally, we conclude and give some perspectives tothis work.Fig.1. Graph extraction approach.2 Graph database Models and Graph AggregationAlgorithm2.1 Graph database ModelsA graph database is defined [13] as a “database where the data structures forthe schema and/or instances are modeled as a (labeled) (directed) graph, orgeneralizationsofthegraphdatastructure,wheredatamanipulationisexpressedbygraph-orientedoperationsandtypeconstructors,andhasintegrityconstraintsappropriate for the graph structure. There is a variety of models for a graphdatabase(formoredetailssee[13]).Allthesemodelshavetheirformalfoundation