Extracting Taxonomies from Data - A Case Study Using Fuzzy Formal Concept Analysis

Taxonomies and, more generally, ontologies, are at the core of the semantic web. In practice, it is rare to find data with meta-data markup in accordance with a full ontology, due to the intensive manual effort involved in the production and maintenance of both the ontology and the data. In many cases, however, data is stored in XML documents or relational tables with implicit taxonomic information such as product type, location, business category, etc. In this work we aim to use methods from formal concept analysis (FCA) to extract such embedded taxonomies, as a starting point for creation of a formal ontology or for further processing of the data. Due to noise, data incompleteness, etc, a soft computing approach is necessary for all but the simplest cases.