A Study on Optimization of Top-k Queries in Relational Databases

Abstract

Efficient execution of top-k queries is increasingly becoming a major challenge for relational database technology. Top-k query has emerged as a key requirement in modern application. In these applications, efficient and adaptive evaluation of top-k queries is an integral part of application semantics. In this paper we discuss a query optimization framework of top-k queries that fully integrates rank-join operators. Ranking queries produce results that are ordered on some computed score. A key property of top-k queries is that, users are interested only in the first k results and not in total ranking of all query results. This property directly impacts the optimization of top-k queries by optimizing for the first k results. Traditionally, most real world database systems offer the feature of first n row optimization. The top-k query is define as- Given a database D of m objects, each of which is characterized by n attributes, a scoring function f, according to which we rank the objects in D, and the number of expected results k. Then a top-k query Q returns the k objects with the highest rank in f. In Top-k query, query define on n attribute a1, a2 ,…, an and relation M in the form of R1 ,R2 ,…,RM that each ai (i=1:n) belongs to one relation Rj (j=1:M). Each of the attributes has special domain in comparison with their kind. According to the query, a series of attributes of these relations are applied for projection, a series of attributes of these relations are used for restriction and join. In the rank aware queries there is apart for ranking that some of relations attribute are presented in the form of a ranking relation which is called ranking function. Ranking function f is formed in the form of attribute n' that is n' <=n. A theory for ranking function f is this: ranking function changes in comparison with all relations are monotonic. In addition to this, the number of suitable answers in rank aware queries is determined too that is just Top k. Consider a set of relations R1 to Rm. Each tuple in Ri is associated with some