Skip to content

Proceedings - NLP 1996

NLP Track at TREC-5

Tomek Strzalkowski, Karen Sparck Jones

Abstract

NLP track has been organized for the first time at TREC-5 to provide a more focused look at how NLP techniques can help in achieving better performance in information retrieval. The intent was to see if NLP techniques available today are mature enough to have an impact on IR, specifically if and when they can offer an advantage over purely quantitative methods. This was also a place to try some more expensive and more risky solutions than those used in main TREC evaluations.

Bibtex 
@inproceedings{DBLP:conf/trec/StrzalkowskiJ96,
    author = {Tomek Strzalkowski and Karen Sparck Jones},
    editor = {Ellen M. Voorhees and Donna K. Harman},
    title = {{NLP} Track at {TREC-5}},
    booktitle = {Proceedings of The Fifth Text REtrieval Conference, {TREC} 1996, Gaithersburg, Maryland, USA, November 20-22, 1996},
    series = {{NIST} Special Publication},
    volume = {500-238},
    publisher = {National Institute of Standards and Technology {(NIST)}},
    year = {1996},
    url = {http://trec.nist.gov/pubs/trec5/papers/nlp.track.96.ps.gz},
    timestamp = {Thu, 12 Mar 2020 00:00:00 +0100},
    biburl = {https://dblp.org/rec/conf/trec/StrzalkowskiJ96.bib},
    bibsource = {dblp computer science bibliography, https://dblp.org}
}

Information Retrieval and Trainable Natural Language Processing

John D. Burger, John S. Aberdeen, David D. Palmer

Abstract

Existing work on indexing and retrieving documents from large on-line collections has had great success at treating both documents and queries as simple, unstructured collections of individual words (terms) with dependencies among these terms largely ignored. However, natural language text has a great deal of structure. In particular, at a scale close to that of the individual word, there are interactions and dependencies that many IR systems ignore. Those systems that do attempt to capture some of these dependencies do so in rather indirect ways. MITRE has substantial experience with trainable natural language algorithms, and we believe this powerful approach is complimentary to the standard information retrieval paradigm. Developed over a number of years, our technology has been used to good effect in the context of information extraction tasks, such as the Message Understanding Conferences (MUC) [1]. Because our approach is based on automatically training our NLP components on large data sets, resulting in accurate and very fast text processors, and because we believe that empirical evaluation is highly important, we have been interested in classical information retrieval for some time. Our first foray into TREC was begun to explore the possibility of improving IR systems with this technology.

Bibtex 
@inproceedings{DBLP:conf/trec/BurgerAP96,
    author = {John D. Burger and John S. Aberdeen and David D. Palmer},
    editor = {Ellen M. Voorhees and Donna K. Harman},
    title = {Information Retrieval and Trainable Natural Language Processing},
    booktitle = {Proceedings of The Fifth Text REtrieval Conference, {TREC} 1996, Gaithersburg, Maryland, USA, November 20-22, 1996},
    series = {{NIST} Special Publication},
    volume = {500-238},
    publisher = {National Institute of Standards and Technology {(NIST)}},
    year = {1996},
    url = {http://trec.nist.gov/pubs/trec5/papers/mitre2.ps.gz},
    timestamp = {Thu, 12 Mar 2020 00:00:00 +0100},
    biburl = {https://dblp.org/rec/conf/trec/BurgerAP96.bib},
    bibsource = {dblp computer science bibliography, https://dblp.org}
}

Xerox TREC-5 Site Report: Routing, Filtering, NLP, and Spanish Tracks

David A. Hull, Gregory Grefenstette, B. Maximilian Schulze, Éric Gaussier, Hinrich Schütze, Jan O. Pedersen

Abstract

Xerox participated in TREC 5 through experiments carried out separately and conjointly at the Rank Xerox Research Centre (RXRC) in Grenoble and the Xerox Palo Alto Research Center The remainder of this report is divided into three sections. The first section describes our work on routing and filtering (Hull, Schütze, and Pedersen), the second section covers the NLP track (Grefenstette, Schulze, and Gaussier), and the final section addresses the Spanish track (Grefenstette, Schulze, and Hull).

Bibtex 
@inproceedings{DBLP:conf/trec/HullGSGSP96,
    author = {David A. Hull and Gregory Grefenstette and B. Maximilian Schulze and {\'{E}}ric Gaussier and Hinrich Sch{\"{u}}tze and Jan O. Pedersen},
    editor = {Ellen M. Voorhees and Donna K. Harman},
    title = {Xerox {TREC-5} Site Report: Routing, Filtering, NLP, and Spanish Tracks},
    booktitle = {Proceedings of The Fifth Text REtrieval Conference, {TREC} 1996, Gaithersburg, Maryland, USA, November 20-22, 1996},
    series = {{NIST} Special Publication},
    volume = {500-238},
    publisher = {National Institute of Standards and Technology {(NIST)}},
    year = {1996},
    url = {http://trec.nist.gov/pubs/trec5/papers/real-xerox.ps.gz},
    timestamp = {Thu, 08 Oct 2020 01:00:00 +0200},
    biburl = {https://dblp.org/rec/conf/trec/HullGSGSP96.bib},
    bibsource = {dblp computer science bibliography, https://dblp.org}
}

Evaluation of Syntactic Phrase Indexing – CLARIT NLP Track Report

Xiang Tong, ChengXiang Zhai, Natasa Milic-Frayling, David A. Evans

Bibtex 
@inproceedings{DBLP:conf/trec/TongZME96b,
    author = {Xiang Tong and ChengXiang Zhai and Natasa Milic{-}Frayling and David A. Evans},
    editor = {Ellen M. Voorhees and Donna K. Harman},
    title = {Evaluation of Syntactic Phrase Indexing -- {CLARIT} {NLP} Track Report},
    booktitle = {Proceedings of The Fifth Text REtrieval Conference, {TREC} 1996, Gaithersburg, Maryland, USA, November 20-22, 1996},
    series = {{NIST} Special Publication},
    volume = {500-238},
    publisher = {National Institute of Standards and Technology {(NIST)}},
    year = {1996},
    url = {http://trec.nist.gov/pubs/trec5/papers/CLARIT-NLP-corrected.ps.gz},
    timestamp = {Thu, 12 Mar 2020 00:00:00 +0100},
    biburl = {https://dblp.org/rec/conf/trec/TongZME96b.bib},
    bibsource = {dblp computer science bibliography, https://dblp.org}
}