The system allows a user to perform automated analysis of any corporate data in order to discover hidden interdependencies. In the first version of the system, data is clustered in order to identify groupings with similar properties.

Typically, an agent is assigned to each record and cluster, if any. Every agent is allowed a certain sum of money, which depends, for example, on the price of the item depicted in the record. Agents of records and clusters negotiate amongst themselves to find the most profitable clustering arrangement, making clustering offers to each other. In accepting or rejecting offers, the agents consider the criteria of the cluster's wealth accumulation and its life expectancy. Records can 'invest' their money into clusters. As a result, clusters can grow larger and richer (if they prove to be useful) or go bankrupt and disappear (if useless). This mechanism enables the self-organisation and evolution of the system of clusters in real time.

The Knowledge Management System is extensively used inside other MagentA systems. For example, in the dialogue management subsystem, inside the MagentA e-Commerce System, it is used to establish patterns of behaviour of site visitors (see above).

Back to MagentA Products

MagentA Home