We are building a Gourmet Advisor system as proof of concept of Complex Adaptive Information Systems. Knowledge management is done using networks of memes, or pieces of information. In this paper we introduce meme networks and describe how a simple learning scheme enables these networks to handle personalization, vague keywords, learning of new knowledge and constant adaptation to usage by users. During interaction with users, parts of a meme network are lifted into a workspace, where the memes interact. This interaction is modeled using activation spreading. During usage, new meme structures emerge which represent the meaning of vague terms and context dependent knowledge, such as personalization.
展开▼