Syntactic Parsing of Web Queries

Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 1787–1796, Austin, Texas, November 1-5, 2016. c⃝2016 Association for Computational Linguistics Syntactic Parsing of Web Queries Xiangyan Sun Fudan University Haixun Wang Facebook Yanghua Xiao∗ Fudan University Zhongyuan Wang Microsoft Research Abstract Syntactic parsing of web queries is impor- tant for query understanding. However, web queries usually do not observe the grammar of a written language, and no labeled syntactic trees for web queries are available. In this pa- per, we focus on a query’s clicked sentence, i.e., a well-formed sentence that i) contains all the tokens of the query, and ii) appears in the query’s top clicked web pages. We ar- gue such sentences are semantically consistent with the query. We introduce algorithms to de- rive a query’s syntactic structure from the de- pendency trees of its clicked sentences. This gives us a web query treebank without manual labeling. We then train a dependency parser on the treebank. Our model achieves much better UAS (0.86) and LAS (0.80) scores than state-of-the-art parsers on web queries. 1 Introduction Syntactic analysis is important in understanding a sentence’s grammatical constituents, parts of speech, syntactic relations, and semantics. In this paper, we are concerned with the syntactic structure of a short text. The challenge is that short texts, for example, web queries, do not observe grammars of written languages (e.g., users often overlook capital- ization, function words, and word order when creat- ∗ Correspondence author. This paper was supported by National Key Basic Reserach Program of China un- der No.2015CB358800, by National NSFC(No.61472085, 61171132, 61033010, U1509213), by Shanghai Municipal Sci- ence and Technology Commission foundation key project under No.15JC1400900. ing a web query), and applying parsers trained on standard treebanks on queries leads to poor results. Syntactic structures are valuable for query under- standing. Consider the following web queries and their syntactic structures we would like to construct: cover iphone 6 plus NN NN CD NN nn num amod distance earth moon NN NN NN nn nn faucet adapter female NN NN ADJ nn nn The syntactic structure of query cover iphone 6 plus tells us that the head token is cover, in- dicating its intent is to shop for the cover of an iphone, instead of iphones. With this knowledge, search engines show ads of iphone covers instead of iphones. For distance earth moon, the head is distance, indicating its intent is to find the dis- tance between the earth and the moon. For faucet adapter female, the intent is to find a female faucet adapter. In summary, correctly identifying the head of a query helps identify its intent, and correctly identifying the modifiers helps rewrite the query (e.g., dropping non-essential modifiers). Syntactic parsing of web queries is challenging for at least two reasons. First, grammatical signals from function words and word order are not avail- able. Query distance earth moon is missing function words between (preposition), and (coordi- nator), and the (determiner) in conveying the intent 1787 distance …

Content truncated to 500 words