Overview of the Third Text REtrieval Conference (TREC-3)
Held in Gaithersburg, MD, August November 2-4, 1994. The conference was co-sponsored by the National Inst. of Standards and Technology (NIST) and the Advanced Research Projects Agency (ARPA) and was attended by 150 people involved in the 32 participating groups. Evaluates new technologies in text retrieval. Includes 34 papers: indexing structures, fragmentation schemes, probabilistic retrieval, latent semantic indexing, interactive document retrieval, and much more. Numerous graphs, tables and charts.
What people are saying - Write a review
We haven't found any reviews in the usual places.
ad-hoc algorithm analysis average precision Boolean Bruce Croft Build in hours Build Query CD CD combination concepts conceptual graph CPU seconds Data Structures database Disk document frequency documents retrieved evaluation Fallout final query G c F Gerard Salton improvement information retrieval INQUERY interactive Inverted index iterations manual hours match median ments N-gram NIST number of documents number of manual number of relevant number of terms performance phrases Query Construction query expansion query formulation query terms R-precision ranking Recall relevance feedback relevance judgments relevant docu relevant documents retrieval system routing queries Run ID scheme scores searchers similarity SMART Statistics stopword Storage in MB Table term frequency Term Weighting Text REtrieval Conference tokens total number training data TREC vector Vector Space Model words
Page 228 - GW and Harshman, RA Indexing by latent semantic analysis. Journal of the Society for Information Science, 1990, 41(6), 391-407.
Page 37 - In Proceedings of the Seventeenth Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 302-310, Dublin, Ireland, 1994.
Page 229 - The term-by-document matrix is decomposed into a set of k, typically 100 to 300, orthogonal factors from which the original matrix can be approximated by linear combination. Instead of representing documents and...
Page 70 - In choosing a term weighting system, low weights should be assigned to high-frequency terms that occur in many documents of a collection, and high weights to terms that are important in particular documents but unimportant in the remainder of the collection. The weight of terms that occur rarely in a collection is relatively unimportant, because such terms contribute little to the needed similarity computation between different texts.
Page 21 - Xerox Palo Alto Research Center, 3333 Coyote Hill Road, Palo Alto, CA 94304.
Page A-10 - R documents (whether relevant or non-relevant) have been retrieved, where R is the number of relevant documents for a topic.
Page 98 - To select the document for rank r, a collection is chosen by rolling a C-faced die that is biased by the number of documents still to be picked from each of the C collections. The next document from that collection is placed at rank r and removed from further consideration. Rankings produced in this way are guaranteed to respect the...
Page 254 - Full Text Retrieval Based on Probabilistic Equations with Coefficients Fitted by logistic Regression", Proceedings of the Second Text Retrieval Conference (TREC-2), NIST publication, 1994, pages 57-66.
All Book Search results »
Practical Digital Libraries: Books, Bytes, and Bucks
Limited preview - 1997