Dynamic pattern discovery using multi-agent technology
George Rzevski, П.О. Скобелев, Igor Minakov, Вольман Семен Игоревич · 2007
Abstract:- A novel method for dynamic pattern discovery based on multi-agent technology has been developed by authors and tested in a number of applications. The key advantages of the new method are: (a) the discovery process is dynamic and adaptive, i.e., it re-clusters data in real time whenever a new data element arrives; (b) the method is capable of finding all clusters without a need for the user to start the process with a hypothesis; (c) records pro-actively search for suitable clusters; (d) users can prescribe what types of clusters they prefer by adjusting microeconomics of the method. The method is particularly suitable for the discovery of patterns of behaviours of website visitors as they browse.