Ad hoc processes have become a big role in modern organizations. Nevertheless, traditional
workflow solutions have problems to support those processes. As a result, there
is the trend moving away from those inflexible processes to collaborative solutions. Here
contributors are able to design or customize existing steps for different activities, with
their domain specific knowledge. Here the problem arises, that these flexible and maybe
unstructured systems lead to a higher effort for recurrent processes, because every step
has to be rethought and created repeatedly for similar cases. Additionally, unskilled users
may be overstrained, because of the lack of guidance. Then, experienced users with their
domain specific knowledge are needed to complete a process successfully. However, these
experienced users are often not involved in the modeling part of those processes, where
templates are made.
In this work concepts are developed to help users with no or less domain knowledge to
handle the modeling part of those processes successfully. Therefore, the implicit knowledge
of the change histories of different cases is used to give recommendations. In a first
step, the entire change history of the SocioCortex server is analyzed for its features. Afterwards
a concept is developed, to prepare the data in a way to make it usable for the
recommender. Out of these prepared data, a recommender is propagated to help users to
create the task related templates.
Finally, a prototypical implementation of these concept is done within the SocioCortex
eco system, to evaluate them.
Keywords: Collaboration, Knowledge organization, Recommender Systems, Intelligent user assistance
The following figure shows the primary process within the CRISP-DM Model: