In order to effectively disseminate task-relevant and process-scope knowledge, knowledge-intensive enterprises adopt knowledge flows to explicitly represent workers' knowledge needs and referencing behavior of codified knowledge during the execution of business tasks. However, due to differences in expertise and experience, individual workers impose varied knowledge needs on the knowledge flows directed by the workflows they participate in. This study proposes a model of workflow-aware knowledge-flow views, i.e. virtual knowledge flows abstracted from workflow-driven knowledge flows, to provide adaptable knowledge granularity. Moreover, a text mining approach is developed to derive knowledge-flow views from codified knowledge objects of knowledge flows, such as documents. Both task knowledge semantics and task execution sequences are utilized to evaluate the degrees of workers' knowledge demands in workflow contexts. Knowledge management systems can thus present different abstracted knowledge flows to diverse workflow participants, and facilitate knowledge sharing and collaboration.