A feature vector "connects" a concept (entity) to a document collection, i.e., a FV is tailored to the
specific terminology used in a particular document collection. FVs are built considering both the
semantics encoded in an ontology and the dominant lexical terminology surrounding the concepts
(entities) in a text corpus. The underlying idea is that a FV reflects both the semantic and linguistic
neighbourhoods of a particular entity. The semantic neighbourhood is computed based on related entities
and direct properties specified in an ontology (or a fragment of an ontology in case of a broad ontology
(Bhatt et al., 2004)), while the linguistic neighbourhood is based on co-location of terms in a document
collection. Therefore, a FV constitutes a rich representation of an entity that is related to the actual
terminology used in a text corpus. Figure 1 shows an illustration of a FV and how it relates to an entity
and a set of documents. For a more formal definition of a FV, the keen reader is referred to (Tomassen &
Strasunskas, 2010).