Sharing Corpora and Tools to Improve Interaction Analysis

Building and Analyzing a Corpus of Contextualized Traces Collected during a Technology Enhanced Teaching Module

“…This approach addresses the problems of (1) heterogeneity (semantics, formats) of collected data and the corresponding learning contexts, (2) strong coupling between data formats and analysis tools, and (3) non-reproducibility of previous analyses. Existing approaches ( [5], [7], [10], [11], [12]) propose a unique representation of the data to be shared and possibly analyzed using a set of available analysis tools. In order to use the services provided by existing sharing platforms, users have to do a preliminary work of conversion to format data according to the chosen representation.…”

“…Building an authentic learning experiment is a complex and timeconsuming task. As a consequence, having access to research data offers more recognition and visibility for owners [6] while it allows analyses' replication and comparison by other researchers [7]. We proposed a new approach called the "Proxy approach" [8] [9] as a base for a platform architecture that allows researchers to share learning traces corpora, without imposing a unique representation of data, and to flexibly use different analysis tools on different corpora depending on analysis needs.…”

The Proxy Model: A New Approach to Sharing and Analyzing Learning Traces Corpora

“…This will increase the usability of analysis tools by facilitating the interoperability between learning environments and analysis tools. The MULCE (Multimodal Learning and teaching Corpora Exchange) project [2] focused on the importance of sharing learning and teaching corpora between researchers. This work puts emphasis on the necessity of contextualising the shared interaction data collected during collective experimental learning situations.…”

“…As in the MULCE project [2], we stress the need to contextualize the interaction traces using data describing the context of the observed learning situation. In fact, our aim being among others to share corpora between the researchers of the TEL community, the availability of interaction traces is not sufficient for understanding the shared data by a researcher who did not participate in the experiment.…”

“…It is constructed by gathering resources used and collected outside the platform during the TEL experiment. It is possible that analytical work had been achieved on the corpus before its (re)construction in the BEATCORP platform, resources used and produced during the analyses must be integrated into the corpus, allowing other researchers to access, verify and enrich them [2]. In order to distinguish between an original corpus, as it was collected and described by a researcher, including analyses achieved on it outside the platform, and the analytical work performed on the corpus within the framework by means of one or several shared analysis tools, we introduce the concept of analysis corpus.…”