Sponsor: ESC Contract No.: FA9721-04-0001

Virtual communities are becoming increasingly popular, particularly on the Internet, as a means for like-minded individuals to pursue common goals. It is a way to access and to share knowledge and information among participants of such communities without physical or hardware constraints. The concept of a community of interest can be supported in a virtual community in order to bring the appropriate parties together and to share their knowledge with each other. The objective of our work is to investigate a community abstraction approach to agent-based knowledge management. We propose a general model that extends the abstraction of an agent, such that it becomes an actor within a knowledge community. In our bottom-up model, agents themselves, software or human, are the members of the virtual knowledge sharing communities. Agents can choose to join, leave, create and destroy communities, and can be member of many communities simultaneously. A prototype design illustrates this approach.

In an efficient Virtual Enterprise (VE), where all the partners, both sending and receiving messages have to lead to acceptable and meaningful agreements, it is necessary to have common standards (an interaction protocol to achieve deals, a language for describing the messages' content and ontology for describing the domain's knowledge). This paper introduces first the ForEV platform, implemented through a Multi-Agent System. This platform facilitates partners' selection automatic process in the context of VE and includes a negotiation protocol through multi-criteria and distributed constraint formalisms, as well as a reinforcement learning algorithm. Then, Ontology-based Services are proposed to be integrated in ForEV architecture in order to help in the VE formation process. These services will make the platform more open, enabling the establishment of the negotiation process between agents with different ontologies although representing the same domain of knowledge. An Ontology-based Services Agent is the responsible for providing the Ontology-based Services and monitoring the whole agents interaction just in time, without needing of a previous and tedious complete ontology mapping process. In our architecture each agent (either market or enterprise) has its own architecture and functionalities (some developer will design and build the ontology with some tool and, later, the agent will access the generated file/database), which implies the heterogeneity of the all Multi-Agent System.

Content-driven and hybrid recommender systems propose products to customers making use of descriptive features and behavioral patterns, likewise. While most approaches exploit classical information retrieval techniques, e.g., nearestneighbor queries in metric spaces, availability and usage of richer semantic meta-information about products may further improve recommendation quality significantly. Massive taxonomies for product classification are coming of age, e.g, the United Nations Standard Products and Services Classification (UNSPSC), as well as proprietary standards, such as Amazon.com’s classification taxonomies for books, DVDs, CDs, and apparel. We exploit suchlike semantic background knowledge in order to leverage powerful inference opportunities for making user profiles, based upon the products these latter customers purchased, more meaningful. Ample empirical analysis, both offline and online, demonstrates our proposal’s superiority over common existing approaches when user information becomes sparse and implicit ratings prevail.