Auto-index Selection for Data Base Using Maximal Frequent Patterns

Subject Areas : Information Technology in Engineering Design (ITED) Journal

Mohammad Hossein Nadimi ¹

1 - Computer group

Received: 2017-06-19 Accepted : 2017-06-19 Published : 2016-02-20

Keywords:

Abstract :

Data access optimization is one of the most important challenges for organizations and competitive businesses therefore selecting useful index is one of the significant techniques to optimize their databases. With the creation of very large databases and consequently need for more advanced query optimizer; database administrator role was not singly enough for finding suitable indexes in database management systems and finding indexes automatically by these systems was considered by researchers in this field. So far several techniques such as data mining techniques are proposed for finding indexes automatically. However there is a challenge in using data mining techniques, whether all indexes have been found are useful and necessary or not? Obviously creating unnecessary indexes is costly in terms of time and memory. For solving the problem, in this research has been proposed an effective way using Maximal frequent Pattern to reduce time for finding indexes automatically. Also in the proposed method using the allocated proper weight to the patterns which have been found, unnecessary indexes are not created. The proposed method like previous works is evaluated on standard data and queries by several experiments, considering the TPC-H benchmark assessment. Test results show that the automatic creation of required indexes using the proposed method compared to the previous method, which makes use of frequent patterns is less time required.