Rule Evaluation Algorithm for Semantic Query Optimisation


ŞAYLI A., Elibol A.

APPLIED MATHEMATICS & INFORMATION SCIENCES, cilt.7, sa.5, ss.1773-1781, 2013 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 7 Sayı: 5
  • Basım Tarihi: 2013
  • Doi Numarası: 10.12785/amis/070515
  • Dergi Adı: APPLIED MATHEMATICS & INFORMATION SCIENCES
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Sayfa Sayıları: ss.1773-1781
  • Yıldız Teknik Üniversitesi Adresli: Evet

Özet

Semantic Query Optimisation (SQO) in Relational Database Management Systems (RDMSs) is a query optimisation approach which uses rules learned from past queries in order to execute new queries more intelligently without accessing database, whenever possible. The approach is composed of several components: Query Representation, Query Optimisation, Automatic Rule Derivation and Rule Maintenance. This paper focused on the query optimisation component. In RDMSs, during the traditional SQO, different alternative queries of a given query can be constructed using matching rule(s) from the rule set, and then its optimiser selects one of the alternatives as an optimum query which will give the same answer set but it can be executed faster than the original query. One of the main problems occurs during this process is to have many matched rules e.g., if the number of the rules is N, the number of the alternative queries is 2(N) - 1. The construction and the optimisation of these alternatives also take time in addition to the execution of the query. In order to overcome this problem, in this paper we propose a new Rule Evaluation Algorithm. The main goal of the algorithm is to evaluate matching rule(s) and select useful/promising rules. And then use selected rules to construct an optimum query. The algorithm can answer the question of the utility of rules in the query optimisation. The system of the approach based on the algorithm has been implemented and its computational results are given. The experimental results show that the algorithm can trim the number of the rules significantly.