Efficient Query Rewrite for Structured Web Queries

TL;DR

This paper presents algorithms for rewriting structured web queries to ensure minimum result retrieval and low latency, addressing the mismatch between user queries and data sources, validated on real-world shopping data.

Contribution

It formalizes the query rewriting problem for structured data, proves NP-hardness, and offers approximation algorithms with empirical validation.

Findings

Algorithms effectively surface minimum results within time constraints.

The problem is NP-hard to solve optimally.

Empirical results demonstrate practical effectiveness on large-scale data.

Abstract

Web search engines and specialized online verticals are increasingly incorporating results from structured data sources to answer semantically rich user queries. For example, the query \WebQuery{Samsung 50 inch led tv} can be answered using information from a table of television data. However, the users are not domain experts and quite often enter values that do not match precisely the underlying data. Samsung makes 46- or 55- inch led tvs, but not 50-inch ones. So a literal execution of the above mentioned query will return zero results. For optimal user experience, a search engine would prefer to return at least a minimum number of results as close to the original query as possible. Furthermore, due to typical fast retrieval speeds in web-search, a search engine query execution is time-bound. In this paper, we address these challenges by proposing algorithms that rewrite the user query in a principled manner, surfacing at least the required number of results while satisfying the low-latency constraint. We formalize these requirements and introduce a general formulation of the problem. We show that under a natural formulation, the problem is NP-Hard to solve optimally, and present approximation algorithms that produce good rewrites. We empirically validate our algorithms on large-scale data obtained from a commercial search engine's shopping vertical.