Abstract: A fuzzy C- means clustering algorithm based on adaptive chaotic particle swarm optimization (ACPSO) is proposed in the thesis. On one hand, interactive procedure based on FCM is replaced by that of PSO; on the other hand, a balanced adaptive .search strategy is embedded so as to accelerate algorithm convergence, improve the capacity of de-noising, enforce local search ability, and escape from local optimization. It will not only solve indeterminacy of Web document in parts, but also obtain stable clustering results quickly.