论文信息 - TNO TREC7 Site Report: SDR and Filtering

TNO TREC7 Site Report: SDR and Filtering

The TNO spoken document retrieval system is based on the ABBOT Large Vocabulary Continuous Speech Recognition (LVCSR) system [Renals 1998] developed by Cambridge University, Sheffield University and SoftSound, and uses word spotting, the TNO Vector Space Engine and fuzzy matching based on phoneme trigrams for indexing and retrieval. We participated in full SDR mode and experimented with several approaches

Wessel Kraaij | David A. van Leeuwen | Rudie Ekkelenkamp

[1] Karen Spärck Jones,et al. Retrieving spoken documents by combining multiple index sources , 1996, SIGIR '96.

[2] J. J. Rocchio,et al. Relevance feedback in information retrieval , 1971 .

[3] Wessel Kraaij,et al. Phoneme based spoken document retrieval , 1998 .

[4] Djoerd Hiemstra,et al. A Linguistically Motivated Probabilistic Model of Information Retrieval , 1998, ECDL.

[5] Alan F. Smeaton,et al. Taiscéalaí: Information Retrieval from an Archive of Spoken Radio News , 1998, ECDL.

[6] Kuldip K. Paliwal,et al. Automatic Speech and Speaker Recognition: Advanced Topics , 1999 .

[7] Peter Schäuble,et al. New techniques for open-vocabulary spoken document retrieval , 1998, SIGIR '98.