Fill-In-The-Functions: Towards Establishing A Workload For Ranking in Web Databases

The emergence of the deep Web databases have given a new connotation to the concept of ranking query results. Earlier approaches for ranking resorted to analyzing frequencies of database values and query logs or establishing user profiles. In contrast, an integrated approach, based on the notion of a similarity model, for supporting useras well as query-dependent ranking has been recently proposed [28]. An important component of this ranking framework is a workload of ranking functions, where each function represents an individual user’s preferences towards the results of a specific query. At the time of answering a query for which no prior ranking function exists, the similarity model can ensure a good quality of ranking only if a ranking function for a very similar user-query pair exists in the workload. Thus, when there exists no function corresponding to a user asking a query, the framework must ensure that the workload contains a ranking function for at least one user-query pair similar to this pair. In this work, we address this problem of establishing an appropriate workload of ranking functions to support userand query-dependent ranking on Web databases. Toward this, we propose a novel metric, termed as Workload Goodness, that determines the appropriateness of a given set of user-query pairs in assisting the Similarity model. We then prove that the optimal solution to this problem is NPcomplete, and propose two distinct algorithms based on greedy approaches for individually determining an appropriate set of user-query pairs. We discuss the effectiveness of our proposal analytically as well as experimentally over two real Web databases.