Sharing Corpora and Tools to Improve Interaction Analysis

A very wide range of online interaction analysis staying in the hands of researchers, and tools being implemented in research prototypes, often used only in non-replicated experimentations, we point out the need for TEL research community to reach large scale validation for its results. This paper is a concrete step in this direction. For a deeper collaboration in our community, we suggest to share structured data collections. The Mulce project aims at proposing a structure for teaching and learning corpora (including pedagogical and research context), and especially for interaction tracks. Two main corpora are built according this structure. This paper defines a teaching and learning corpus, shows its main structure and browses some parts of the structured interaction data. We also describe the platform that enables the community to browse and analyze a shared corpus.

• Type: Conference
• Year: 2009
• Where: ECTEL
• Authors: Christophe Reffay, Marie-Laure Betbeder
• Keywords: ECTEL 2009, Machine Learning, Online Interaction Analysis, Structured Interaction Data, TEL Research Community