Exploring The Most Frequent Heuristic for (Semi)-Structured Retrieval
Arash Termehchy, Adarsh Prasad, Abraham Engle, Marainne Winslett · Illinois Digital Environment for Access to Learning and Scholarship (University of Illinois at Urbana-Champaign) · 2011
Keyword query interfaces for databases (K QIs) provide easy access to data, but often suffer from ranking quality problems.Since users do not generally specify the schema elements in keyword queries, an important challenge to build an effective KQI is to find the desired schema elements for a given keyword query.One popular heuristic, called the most frequent heuristic, assumes that the schema elements where the query terms appear more frequently are the desired schema elements for the query terms.In this paper, we have performed an extensive empirical evaluation using keyword queries form a Web search engine.We show that the most frequent heuristic is not generally effective and in most cases users prefer the schema elements where the query terms appear quiet infrequently.