Architecture of collaboratively-changeable hierarchical structure

A new approach under development to collaborative navigation on a large collection of heterogeneous resources, to be carried out by a large number of users. An important difference from conventional Crowdsourcing is the complete decentralization of management.
The approach is based on the selection and bringing together income of like-minded users to the classification of the collection. Thinking outside the box a user spends adequate efforts on building individual taxonomy, accurate within its immediate interests. The more active user in the interface and the more he has active in same contexts like-minded persons, the more faster and deeply taxonomy will be adjusted for his mind.
The robots that analyze the resources can participate among the users.
The general architecture of the future system and a few algorithms are presented in the paper.