Transmute : un outil interactif pour assister l'extraction de connaissances à partir de traces
Abstract
Analysis of traces is an increasingly active issue in knowledge discovery (kd). Unfortunately,
to date, few tools are designed to help interactively the analyst during the
knowledge discovery process. In this paper, we present Transmute, a tool to assist
the analyst throughout an interactive kd process in traces. Transmute displays the
results of the mining together with the source trace so that it is easier for the analyst
to understand them and thus to refine the parameters of the algorithm accordingly.
Transmute is developed as a web platform and relies on a Python implementation of
the kd process.