Search Strategies for Keyword-based Queries

Given a set of keywords, we find a maximum Web query (containing the most keywords possible) that respects user-defined bounds on the number of returned hits. We assume a real-world setting where the user is not given direct access to a Web search engine´s index, i.e., querying is possible only through an interface. The goal to be optimized is the overall number of submitted Web queries. One original contribution of our research is the formalization and theoretical foundation of the problem. But, in particular, we develop a co-occurrence probability informed search strategy for the problem. The performance gain achieved with our approach is substantial: compared to the uninformed baseline (without co-occurrence information) the expected savings are up to 20% in the number of submitted queries and runtime.