Identifying Media-Specific and Time-Dependent Patterns in Community Success Models

Community Information Systems support professional communities of practice in organizing their work and learning processes on the Web. In such systems, communities draw upon an ecosystem of multimedia technologies to build and manage knowledge, following a cross-medial transcriptivity theory, as indicated in the figure above. Drawing upon a pool of existing media artifacts (prescripts), communities select tools and algorithms (scripts) to transcribe combinations of existing prescripts to new media artifacts (transcripts), addressed to and localized for the respective community. These new media artifacts become potential prescripts for further transcription operations. With such cross-media transcriptions, communities build new knowledge, addressed to and localized for their respective community context. We assume that communities use identifiable media-specific and time-dependent patterns of these transcription processes to reach their goals. Moreover, we assume that best practices for organizing work and learning emerge from the visual analytics and comparison of real communities. The goal of this thesis is to analyze transcription processes from real communities using data mining and machine learning. As foundation, we provide the MobSOS (Mobile Oracle for Success) framework for community information systems success awareness, including a service tool kit based on las2peer (https://las2peer.org) as well as several MobSOS data sets.