Dynamic Guessing Work for Crucial Access of Keyword Objections over the Databases

Abstract

Keyword queries on databases offer easy get right of entry to facts, however frequently suffer from low ranking high-quality, i.e., low precision and/or do not forget, as shown in recent benchmarks. It would be useful to pick out queries which are probable to have low rating fine to enhance the user pleasure. As an example, the system may additionally recommend to the person alternative queries for such difficult queries. In this paper, we examine the traits of tough queries and propose a novel framework to measure the diploma of trouble for a key-word question over a database, considering both the structure and the content material of the database and the query results. We compare our query difficulty prediction version in opposition to two effectiveness benchmarks for popular key-word seek ranking strategies. Our empirical results show that our version predicts the hard queries with excessive accuracy. Similarly, we gift a suite of optimizations to limit the incurred time overhead.