Many researchers have taken an interest in a collective intelligence the development of the user-generated web. Knowledge publishing and management with interactions between users are mostly a social revolution from the wisdom of crowds. This paper proposes a tag based collaborative knowledge management system. It can be realized both the collective intelligence with the crowdsourcing system and the dynamic categorization with the collaborative tagging approach. In addition, we establish the trustworthy knowledge through significant role allotment between a user group and a knowledge expert group.