Ontology has been widely adopted as the basis of knowledge sharing and knowledge-based public services. However, ontology construction is a big challenge, especially in collaborative ontology development, in which conflicts are often a problem. Traditional collaborative methods are suitable for centralized teamwork only, and are ineffective if the ontology is developed and maintained by mass broadly distributed participators lacking communications. In this kind of highly collaborative ontology development, automated conflicts detection is essential. In this paper, we propose an approach to classify and detect collaborative conflicts according to some mechanisms: 1) impact range of a revision, 2) semantic rules, and 3) heuristic similarity measures. Also we present a high effective detecting algorithm with evaluation.