We present our three main results in adressing the problem of assisting the socio-cognitive analysis of human interaction. First, we propose a description of the process of analysis of such data, as well as a generic artefact which covers a large number of the analytic artefacts we have observed and which we call a replayable. Second, we present a study and a modelling of replayables, and describe the four fundamental operations which can be applied to them: synchronisation, visualisation, transformation and enrichment. Finally, we describe the implementation of this model in an environment that assists analysis through the manipulation of replayables, which we evaluate in real-life research situations. Tatiana (http://code.google.com/p/tatiana), the resulting software environment, is based on these four operations and integrates numerous possibilities for extending these operations to adapt to new kinds of analysis while staying within the analytic framework afforded by replayables.