Exposing Chat Features through Analysis of Uptake between Contributions

Understanding distributed learning and knowledge creation requires multi-level analysis of local activity and of how this local activity gives rise to larger phenomena in a network. Computational support is needed for such analyses due to the size of the data and distributed nature of interaction. This paper reports on one step towards implementing an analytic framework that addresses these needs. Contingencies, defined as observed relationships between contributions that evidence contextual relevance, are computed according to automatable rules, and combined to infer uptake relations between contributions. The resulting uptake structure is then analyzed through various network-analytic methods and is also transformed into a graph of uptake between actors for social network analysis. Our initial results show that a simple contingency analysis based on temporal factors, actor addressing, and lexical overlap provides structures of sufficient quality for identification of major features of a discussion and the roles of actors. The results are expected to improve as semantic analysis is added.