Causal relation detection for activities from heterogeneous sources

Abstract

On the web, information representing specific activities is often scattered over different systems. Although, causal relations exist between these activities, these are usually not obviously visible to the user, unless explicitly given. This paper outlines the difficulties which are caused by missing relations. The core contribution of this work will be a system which is capable of identifying cause-effect relations between single activities. The system will use these relations to form coarse-grained groups consisting of sequences with single activities. The intended goal is to employ the detected relations to reduce information overload while increasing accountability, clarity, and traceability for its users. The research is conceived under the assumption of handling heterogeneous sources of information. A further objective is to create a highly generic and flexible system which can be adapted to different use cases. The system will be evaluated with concrete case studies, one of them analyzing relations on software development sites such as SourceForge. © 2012 Springer-Verlag.