Contextualized versus Structural Overlapping Communities in Social Media

Overlapping community structures are fundamental building blocks of Question Answer Forums (QAF) which are extremely linked with every social platform. Keeping in mind the value of structural-based Overlapping Community Detection (OCD) techniques, our imperceptible knowledge of content and contextualized information persuades us to do further research in this direction. To investigate community content, we devise two simple OCD algorithms taking use of posts in forums. In addition, we crawl data of an open source software project on modelling of chemical structures and make it available online for additional analysis. Furthermore, we analyse the devised methods and baseline algorithms on this dataset over the periods of its releases. Not only content could improve the performance of structural techniques but also two innovative algorithms performed competitive in comparison to other methods. Results also indicate reverse correlation between modularity and similarity of content in this forum, which show modularity and content as significant analytic indicators. Our content-based techniques and investigations can be applied by researchers in analytic and recommender systems.