In task-based business environments, effective knowledge management relies on providing task-relevant information to fulfill the information needs of knowledge workers. This work proposes an adaptive task-based profiling approach to model workers' task needs, namely information needs (profiles) on tasks. Meanwhile, a fuzzy analytical method is proposed to identify peer-groups with similar task-needs based on workers' profiles. Accordingly, a Knowledge Support (K-Support) system is developed to provide a collaborative task-based workplace facilitating knowledge retrieval and sharing among peer-groups. The proposed K-Support system is grounded in a research institute to stimulate task-based knowledge retrieval and sharing. Experimental results show that the proposed system can support task-relevant knowledge effectively. © 2005 Elsevier Ltd. All rights reserved.