An efficient algorithm to generate search shortcuts

This paper proposes an efficient and effective solution to the search shortcut problem. Basically, we aim at suggesting queries to web search engine users in order to help them in rapidly satisfying their information needs. By introducing a weak function for assessing query similarity in the search shortcut model, we resort to a simple technique consisting in processing a full-text query over an inverted file index built over a collection of virtual documents representing user sessions. The query suggested are of high quality and are not affected by the data-sparsity problem. We also pro- pose an objective and reproducible way to assess the quality of a given query suggestion technique based on third-party, publicly-available, and human-assessed data. The accurate experimental evaluation conducted shows that our proposal remarkably outperforms other state-of-the-art solutions.