Cost Optimized Data Access with Rank-Join

Sudhir G. Chavan, Vijay B. Patil

Abstract: The prime task of search computing is to join the result of complex query plans. Join of complex query plan problem is classified in the conventional rank aggregation i. e. combining different ranked lists of objects to produce single valid ranking. Rank-join algorithms provide best overall results without accessing total objects in list. This paper describes further views on topic by emphasizing the study and experiments on algorithms that operate with joining the ranked results produced by search services. The rank-join problem is considered to be extending rank aggregation algorithm to the case of join in setting of relational database. On the other hand search computing join diverges from orthodox relational concepts in many ways. Random and sorted access patterns are used to access the services, accessing service is costly in terms of response time, because usually they are remotely located. The output is returned in pages of answers and criteria is some top-k ranking function, multiple search services to answer the same query, user can also redefine the search criteria. This paper proposes Cost Aware Rank-Join with Random and Sorted Access (CARS) methodology in the context of rank join algorithms for the efficiency of search computing. Experimental results prove that CARS strategy outperforms the existing methods of Data Access in terms of access cost.

Keywords: Top-k, Rank-Join, Cost Optimization, Query Optimization, Search Computing, TA Algorithm