The ability to find online communities with shared expertise or interests is a challenging and interesting problem. The problem involves collecting entity data, at scale, from a number of sources; linking data that refer to the same real-world entity across the sources; enriching the data to add the attributes that may define the community (such as interests); and finally analyzing the resulting data (for example looking at the social connectivity). Compounding these problems is the fact that in many cases the online community of interest is not well defined ahead of time, but instead only becomes crystalized through repeated cycles of data collection, refinement, expansion and analysis (what we call "interactive" sessions). In this paper we present PSI4, an end-to-end system that addresses these issues, and we walk through an interactive session to highlight its capabilities for community finding.
展开▼