| Abstract |
Improving a search system for large audiovisual archives can be done in two ways: by enriching the annotations, or by enriching the query mechanism. Both operations possibly benefit from a preliminary
terminological enrichment of the controlled vocabulary in use, i.e. the thesaurus. In this paper we report on a four-parts experiment in which we evaluate the benefits and drawbacks of both aspects: the added value and pitfalls of automatically generated semantic annotations over classically (i.e. manually) assigned keywords and the added value and pitfalls of query expansion over pure keyword matching technique; we then investigate the combination of these operations in the following setup: we
create the baseline for our experiments by querying a set of documents annotated by cataloguers with keywords from the thesaurus. We then apply the same querying process on a set of annotations automatically generated from textual resources related to the documents. Thirdly, we apply a querying process enhanced with query expansion functionalities to the first set of manually annotated documents. Finally, we apply the query expansion mechanism on the automatically generated annotations.
The results give insight into the interaction between the two approaches.
|