Reconsideration of the Effectiveness on Extracting Computer Diagnostic Rules by Automatically Defined Groups

  • Authors:

  • Yoshiaki Kurosawa;Akira Hara;Kazuya Mera;Takumi Ichimura

  • Affiliations:

  • Graduate School of Information Sciences, Hiroshima City University, 3-4-1, Ozuka-higashi, Asaminami-ku, Hiroshima, Japan;Graduate School of Information Sciences, Hiroshima City University, 3-4-1, Ozuka-higashi, Asaminami-ku, Hiroshima, Japan;Graduate School of Information Sciences, Hiroshima City University, 3-4-1, Ozuka-higashi, Asaminami-ku, Hiroshima, Japan;Graduate School of Information Sciences, Hiroshima City University, 3-4-1, Ozuka-higashi, Asaminami-ku, Hiroshima, Japan

  • Venue:
  • KES '07 Knowledge-Based Intelligent Information and Engineering Systems and the XVII Italian Workshop on Neural Networks on Proceedings of the 11th International Conference
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

Our aim is to manage computer systems without expert knowledge. We have proposed a method of diagnostic rule extraction from log files by using Automatically Defined Groups (ADG) based on Genetic Programming. However, this work less explained the effectiveness, especially, the characteristics of the acquired rules. Therefore, we re-evaluated the effectiveness by performing two experiments: the use of artificial log files and the use of real log files. As a result, we confirmed that ADG could acquire the rules composed of multiple terms. This characteristic is very important because we can judge the message that we must consider the co-occurrence of the words, i.e. `Error' and `not'. Thus, we conclude that the ADG is effective for the diagnosis of the systems.