In this paper, we propose a information filtering method using real-coded genetic algorithm (Real-Coded GA). This method creates user's profile with Real-Coded GA using unimodal normal distribution crossover. The similarity is defined as the inner product between user 'profile and each document vector. When the inner product value is large, this method recommends the document. We compare this method with Linear Programming and Relevance Feedback, and discuss the efficiency of this method.
展开▼