Classical process discovery approaches usually investigate logs generated by processes in order to mine and discovery corresponding process schemas. When the collaboration processes case is addressed, such approaches turn to be poorly effective, due to the fact that: (i) logs of collaboration processes are usually stored in heterogeneous data storages which also expose different data types; (ii) it is not easy and direct to derive a common analysis model from such logs. As a consequence, classical methodologies usually fail. In order to fulfill this gap, in this paper we describe a composite methodology that combines semantics-based techniques and multidimensional analysis paradigms to support effective and efficient collaboration process discovery from log data.
