Information resource recommendation in knowledge processes

This paper proposes a framework and an implementation for pro-active just-in-time information resource delivery, based on knowledge processes. We focus on providing a desktop application that presents a ranked list of information resources, such as documents, web sites or e-mail messages, that are considered to be most relevant in that point in time. The paper decomposes the recommendation problam into subproblems and provides evaluation on several event, action and process models. Results show that defining actions based on clustering of events yields the best recommendations.