A Case Study in Natural Language Based Web Search - Natural Language Processing and Text Mining - page 87

5.5 Conclusion
We deployed a natural language based search to a community of Web users, and
measured its popularity relative to conventional keyword search. Our work addressed
criticisms of NLP approaches to search to the effect that they are not scalable and are
too complex to be usable by average end-users. Our approach rests on a sophisticated
index parameterization of text content, that captures syntactic and semantic roles,
in addition to keyword counts, and enables interactive search and retrieval of events
patterns based on a combination of keyword distributions and natural language
attributes. Our distributed indexing and search services are designed to scale to
large document collections and large numbers of users. We successfully deployed on
a Web site that serves a community of 100,000 users. An analysis of query logs shows
that, during the first six months of operation, tra c has increased by almost 40%.
Even more significantly, we are encountering some success in promoting natural
language searches. Our study demonstrates that the percentage of users that avail
themselves of guided fact navigation based on natural language understanding has
increased from 4% to 10% during the first six months of operation. Going forward,
we will focus on increasing this percentage with a more innovative UI.
5.6 Acknowledgments
This work was partially supported by Dr. Joseph Psotka of the US Army Research
Institute under contract No. W74V8H-05-C-0016. We are also indebted to John
Pike, director of GlobalSecurity.org and his staff for providing us with many months
of user tra c Web logs prior to going live.
References
1. D. Appelt and D. Israel. Introduction to information extraction technology.
IJCAI-99 tutorial. http://www.ai.sri.com/ ∼ appelt/ie-tutorial/ijcai99.pdf.
2. D. Appelt and D. Israel. Semantic approaches to binding theory. In Proceedings
of the Workshop on Semantic Approaches to Binding Theory. ESSLLI , 2003.
3. A. Arampatzis, T. van der Weide, P. van Bommel, and C. Koster. Linguistically-
motivated information retrieval. In M. Dekker, editor, Encyclopedia of Library
and Information Science , Springer Verlag, volume 69, pages 201-222. 2000.
4. C. F. Baker, C. J. Fillmore, and J. B. Lowe. The Berkeley FrameNet project.
In C. Boitet and P. Whitelock, editors, Proceedings of the Thirty-Sixth Annual
Meeting of the Association for Computational Linguistics and Seventeenth Inter-
national Conference on Computational Linguistics , pages 86-90, San Francisco,
California, 1998. Morgan Kaufmann Publishers.
5. I. Dagan, O. Glickman, and B. Magnini. The pascal recognizing textual entail-
ment challenge. In Proceedings of the PASCAL Challenges Workshop Recogniz-
ing Textual Entailment , 2005.
6. M. Dimitrov. A light-weight approach to coreference resolution for named enti-
ties in text. Master's thesis, University of Sofia, 2002.