Providing alternative declarative descriptions for entity sets using parallel concept lattices

Abstract

We propose an approach for modifying a declarative description of a set of entities (e.g., a SPARQL query) for the purpose of finding alternative declarative descriptions for the entities. Such a shift in representation can help to get new insights into the data, to discover related attributes, or to find a more concise description of the entities of interest. Allowing the alternative descriptions furthermore to be close approximations of the original entity set leads to more flexibility in finding such insights. Our approach is based on the construction of parallel formal concept lattices over different sets of attributes for the same entities. Between the formal concepts in the parallel lattices, we define mappings which constitute approximations of the extent of the concepts. In this paper, we formalise the idea of two types of mappings between parallel concept lattices, provide an implementation of these mappings and evaluate their ability to find alternative descriptions in a scenario of several real-world RDF data sets. In this scenario we use descriptions for entities based on RDF classes and seek for alternative representations based on properties associated with the entities.