The purpose of this presentation is to outline the requirements for natural language processing (NLP) when using knowledge organisation systems (KOS) for query expansion (QE). Knowledge organisation systems such as Term lists (authority files, glossaries, dictionaries, gazetteers), Classification and Categories (classification scheme, taxonomy, subject headings) and Relationship lists (thesaurus, semantic network, ontology) are common tools for query expansion. Each type of KOS carries different characteristics such as the type and depth of hierarchy, the relationships of terms, the domain specificity, the way of their development (manual, automatic or semi-automatic) to name a few. In order to match natural language query terms to terms from KOS, these characteristics have to be taken into consideration because they define the required processing of variant approaches in query expansion techniques. The issues in question extend from matching words at a lexical level to high-level concept mapping.
Talk slides in pdf [~1.13MB]http://www.iit.demokritos.gr/docs/seminars/mastora_20120208.pdf