Search Logs Mining: Survey

Abstract

Search engine process millions of query and collect data of user interaction every day. These huge amount of data contains valuable information through which web search engine can be optimized. Search engine mostly relies on explicit judgement received from domain experts. To survive the competition search engine must understand user’s information needs very well. Search logs provide implicit data about user’s interaction with search engine. Search logs are noisy, they contain data of both successful search and unsuccessful search. The challenge is to accurately interpret user’s feedback to search engine and learning the user access patterns, such that search engine will better be able to cater the user’s information needs. User feedback can be used to re-rank the search result, query suggestion and URL recommendation.