ZURÜCK Associative Indexing and Query Expansion

In an experimental setting we used co-occurrence data for automated indexing (Ferber 1997): [->] A corpus of bibliographic records intellectually indexed with the OECD thesaurus was used to extract similarities between the words of the titles and the descriptors of the thesaurus. Two sets of records were kept aside: one to optimize the parameters of the system, the other to test it. The word - descriptor similarities were used to predict the intelectually given descriptors of the test records.


© 1998 / HTML-Version 17. 11. 1998: R. Ferber