Beauty is in the eye of the beholder. In
(Strasunskas & Tomassen, 2008) and (Tomassen &
Strasunskas, 2009) end-users were used to assess
the performance of our approach to semantic
search. Therefore, FV quality was measured
indirectly. In this paper, we directly evaluate
quality of FVs relative to the ontologies used on
the Web. Ideally, Text Retrieval Conference
(TREC) corpus would be used but we experienced
the same problems as d'Aquin et al. (2008) in
finding good ontologies that covered TREC
documents and queries. Therefore, we proposed
the following intrinsic and extrinsic measures to
evaluate the quality of the FVs. The Average Fv
Similarity (AFvS) and the Average Fv
Neighbourhood Similarity (AFvNS) are both
intrinsic measures that indicate the FV quality
w.r.t. the ontology (the latter assesses the semantic
neighbourhood of the entities). Further, the
Average Fv NGD (AFvNGD) is an extrinsic
measure that indicates the FV quality w.r.t. the
Web. Finally, the Fv Quality Score (FvQS)
aggregates the overall FV quality score. Next, the
different scores used to measure the FV quality are
defined.