LGMD: Optimal Lightweight Metadata Model for Indexing Learning Games

International audience; Learning Games (LGs) have proven to be effective in a large variety of academic fields and for all levels; from kindergarten to professional training. They are therefore very valuable learning resources that should be shared and reused. However, the lack of catalogues that allow teachers to find existing LGs is a significant obstacle to their use in class. It is difficult for catalogues, or any type of search engine, to index LGs because they are poorly referenced. Yet, many researches have proposed elaborate metadata models for LGs. However, all these models are extensions of LOM, a metadata model that is widely used for referencing learning resources, but that contains more than 60 fields, of which more than half are irrelevant to LGs. The gap between these models and the information that game designers are willing to provide is huge. In this paper, we analyze the LG metadata models proposed in previous research to detect the fields that are specific to LGs and the fields that are irrelevant to LGs. We then propose LGMD (Learning Games Metadata Definition), an optimal lightweight metadata model that only contains the important information for LG indexing. LGMD reduces by two thirds the number of fields compared to the previous models. We confronted this model with the information actually provided by LG editors , by analyzing 736 LG page descriptions found online. This study shows that LGMD covers all the information provided by the LG editors.