2008 | OriginalPaper | Buchkapitel
An Entry Vocabulary Module for a Political Science Test Collection
verfasst von : Benjamin Berghaus, Thomas Mandl, Christa Womser-Hacker, Michael Kluck
Erschienen in: Business Information Systems
Verlag: Springer Berlin Heidelberg
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
We describe the design of a retrieval test for texts on political science. The corpus contains 600,000 documents in various languages. A set of 25 typical topics for the domain was developed and relevance judgments were provided by domain experts. To improve the domain specific retrieval performance, an entry vocabulary module (EVM) which maps query terms to the domain specific vocabulary was developed. We compare a base run to a blind relevance, feedback run as well as to both a static and a dynamic EVM. The dynamic EVM is presented in this paper. It can be shown that the dynamic EVM greatly improves recall and also improves precision. An innovative topic specific analysis proves that the EVM also hurts some topics.