Knowledge cores in large formal contexts (Q2149803)

This paper deals with the problem of extracting useful knowledge from large relational data sets in the context of formal concept analysis (FCA). The problem here is that, on the one hand, extracting meaningful knowledge characterizing the given data set is computationally hard and, on the other hand, the approaches aimed at reducing the size of the data sets to be analysed (so as to make knowledge extraction computationally feasible) bring the indirect negative consequences of either returning patterns that are hardly interpretable or limiting discovered knowledge to highly supported patterns, thus possibly missing interesting but rare ones. This paper proposes a new approach based on exploiting the concept of networks' \(k\)-cores. In particular, the paper first introduces the notion of \(k\)-core in the context of formal concept analysis and then uses the thus defined formal gadget to obtain techniques for knowledge extraction from large relational data sets. Experiments are also provided to demonstrate the validity of the presented approach.