The Linguistic Structure of English Web-Search Queries

Proceedings of the 2008 Conference on Empirical Methods in Natural Language Processing, pages 1021–1030, Honolulu, October 2008. c⃝2008 Association for Computational Linguistics The Linguistic Structure of English Web-Search Queries Cory Barr and Rosie Jones Yahoo! Inc. 701 First Ave Sunnyvale, CA 94089 barrc,jonesr@yahoo-inc.com Moira Regelson Perfect Market, Inc. Pasadena, CA 91103 mregelson@perfectmarket.com Abstract Web-search queries are known to be short, but little else is known about their structure. In this paper we investigate the applicability of part-of-speech tagging to typical English- language web search-engine queries and the potential value of these tags for improving search results. We begin by identifying a set of part-of-speech tags suitable for search queries and quantifying their occurrence. We find that proper-nouns constitute 40% of query terms, and proper nouns and nouns together constitute over 70% of query terms. We also show that the majority of queries are noun- phrases, not unstructured collections of terms. We then use a set of queries manually la- beled with these tags to train a Brill tag- ger and evaluate its performance. In addi- tion, we investigate classification of search queries into grammatical classes based on the syntax of part-of-speech tag sequences. We also conduct preliminary investigative experi- ments into the practical applicability of lever- aging query-trained part-of-speech taggers for information-retrieval tasks. In particular, we show that part-of-speech information can be a significant feature in machine-learned search- result relevance. These experiments also in- clude the potential use of the tagger in se- lecting words for omission or substitution in query reformulation, actions which can im- prove recall. We conclude that training a part- of-speech tagger on labeled corpora of queries significantly outperforms taggers based on tra- ditional corpora, and leveraging the unique linguistic structure of web-search queries can improve search experience. 1 Introduction Web-search queries are widely acknowledged to be short (2.8 words (Spink et al., 2002)) and to be fre- quently reformulated, but little else is understood about their grammatical structure. Since search queries are a fundamental part of the information retrieval task, it is essential that we interpret them correctly. However, the variable forms queries take complicate interpretation significantly. We hypoth- esize that elucidating the grammatical structure of search queries would be highly beneficial for the as- sociated information retrieval task. Previous work with queries (Allan and Raghavan, 2002) considered that short queries may be ambigu- ous in their part of speech and that different docu- ments are relevant depending on how this ambigu- ity is resolved. For example, the word “boat” in a query may be intended as subject of a verb, object of a verb, or as a verb, with each case reflecting a distinct intent. To distinguish between the possi- bilities, Allan and Raghavan (Allan and Raghavan, 2002) propose eliciting feedback from the user by showing them possible contexts for the query terms. In addition to disambiguating query terms for re- trieval of suitable documents, part-of-speech tag- ging can help increase recall by facilitating query reformulation. Zukerman and Raskutti (Zukerman and Raskutti, 2002) part-of-speech …

Content truncated to 500 words