An Experimental Application of Formal Concept Analysis to Research Communities

Formal concept analysis (FCA) is a tool that automatically converts relationships between two domains into an ontology for study. While powerful, this technique sometimes falls short due to its inability to handle sparsely populated relations. This is demonstrated when using FCA to identify research communities merely from collections of coauthorship data. To avoid the pitfalls caused by such datasets, a new technique is built on FCA's foundations: power concept analysis. This expansion allows for the grouping of one domain into equivalence classes, thereby sharing relationships without altering the overall flavor of the context and maintaining correctness of traditional FCA techniques. Taking advantage of this additional exibility, tools are utilized from social network analysis to identify collaboration clusters that better model real-world phenomena. This complete technique is referred to as social network concept analysis, or SNCA, and is leveraged to map research communities to formal concepts.