Blind Men and the Elephant: Detecting Evolving Groups In Social News

• DOI: 10.48550/arxiv.1304.1567
• Publication Date: 2013-01-01
• Authors: Roja Bandari; Hazhir Rahmandad; Vwani Roychowdhury
• Keywords: Representation; Ground truth

We propose an automated and unsupervised methodology for a novel summarization of group behavior based on content preference. show that graph theoretical community evolution (based similarity user preference content) is effective in indexing these dynamics. Combined with text analysis targets automatically-identified representative each community, our method produces multi-layered representation evolving behavior. demonstrate this the context political discourse social news site data spans more than four years find coexisting leanings over extended periods disruptive external event lead to significant reorganization existing patterns. Finally, where there exists no ground truth, we new evaluation approach by using entropy measures as evidence coherence along path groups. This valuable designers managers online forums need granular analytics activity, well researchers sciences who wish extend their inquiries large-scale available web.