Log-mining-based query spelling correction for Chinese search engines

Bo Zhou,Min Zhang,Shaoping Ma,Yiqun Liu,Liyun Ru
2009-01-01
Journal of Computational Information Systems
Abstract:Query spelling correction is crucial to search engines for improving web search relevance and user experience. Additional difficulty is posed beyond traditional lexicon-based spelling correction, since there exist vast amount of out-of-lexicon but legitimate terms in queries. Furthermore, context sensitive techniques cannot work due to the short query fact. Most recent studies concentrate on spelling correction of English queries. Nevertheless, in Chinese search engines, queries contain more mixed types of terms and more complicated spelling error patterns as compared to those in English search engines. In this paper, we analyze the types and patterns of query spelling errors in Chinese search engines. By using language-dependent error models, we present a query spelling correction approach for Chinese search engines, in which knowledge extracted from search log is embedded. The experiments performed on randomly sampled real search queries show that % precision and % recall have been achieved. 1553-9105/ Copyright © 2009 Binary Information Press.
What problem does this paper attempt to address?