چكيده به لاتين
Abstract:
Due to the ever-growing increase in the amount of data we deal with, tapping methods like data mining to extract hidden knowledge and information in the data seems inevitable. One of the topics of data mining that has recently attracted a lot of attention is the inherent distribution of data. The development of computer network technology and distributed database technology has promoted distributed data storage and the new technical generation of distributed data mining. Distributed data mining (DDM) uses distributed computing and finds the required knowledge for the users from the distributed database. This domain has widespread applications. The purpose of data mining from distributed information systems is usually threefold: 1) Identifying locally significant patterns in individual databases; 2) Combining local patterns and discovering global patterns after unifying distributed databases in a single view; and 3) Finding patterns which follow special relationships across different data collections.
Considering agent and multi-agent capabilities in distributed environments, it seems that using their features can be useful in these environments. In this study, in addition to reviewing the related work and researches in agent-based data mining area, we intend to consider the problem of mining association rules in distributed environments. We study this problem in two phases. First we mention the features and capabilities of agents for data mining task, and analyze the advantages of multi-agent combining with distributed data mining. Then, in the second phase we propose a DDM architecture based on multi-agent technology. To conclude, the main result of this thesis is the presentation of agent-based approach for DDM with concept drift using goal-oriented, intelligence, learning and reasoning features of agents.
Keywords: Distributed data mining, Muli-Agent system, Association rules,Agent
