Monitoring learning activities using social knowledge

Learning activities use an increasing number of software tools. These mediate the interactions among participants and of these with resources like documents and tools. These tools constitute a valuable source of information about the actual learning processes. However, this use faces multiple problems. Lecturers need to gain expertise on the data each tool provides and how to analyse them, each tool only offers a partial view of the process, students have different profiles and use different tools, and time to make analyses is limited. To address this situation, this work proposes the use of Assistants for Learning Activities (ALAs), i.e. semi-automated tools that use social knowledge to integrate different sources of information and interpret their data. This knowledge is extracted from literature on learning, and specified as social properties. These properties describe patterns that appear in information and their interpretation in terms of the learning-related activities. Their specification relies on a specific modelling language oriented to social activities and their context. It is designed to facilitate communication with the target learning communities. Wrappers for software tools get the raw data, transform them into facts for this language, and assert them in an information base. Then, a pattern matching algorithm finds instances of the social properties among these facts, giving an interpretation of the original data. A case study on teamwork in a project-based learning context of a university using several software tools illustrates the approach. It shows the feasibility of adapting the analysis through the modification of the considered properties, and how these can explain the observed data.