Abstract:

The use of technology in everyday tasks offers the possibility of collecting large amounts of observations of events in different environments. Most tools can store a detailed account of user activities in specific files commonly referred to as ‘’logs’’. These files can be analyzed to deduce information that is not directly visible, such as which resources a user uses the most, how long a user has been online, how many times a user has interacted in a forum, etc. Graphical visualizations of this data, most often in histograms or bar graphs, line graphs or scatter plots, can be used to support the analysis and monitoring of activities. Visualization can also be applied in the educational domain to evaluate the learner activities from data collected by the learning environment.

Several tools have been proposed in learning environments such as, for example, learning analytics dashboards support learners self-regulation and awareness and assist teachers in online or offline teaching. Most of these dashboards are deployed to help teachers better understand learner activity in a course, reflect on their teaching practice, and find learners at risk of drop out or failure. As for learners, a learning analytics dashboard is there to help them analyze, reflect and make decisions to improve, change or keep their learning methods by putting them in front of their activities and use. To achieve this, however, learners must find the represented data helpful, the visualizations clear and understandable, and easily transform their reflection into action. However, several researchers have deplored the low level of use of learning analytics dashboard by learners, perhaps because they did not find them helpful.

We, therefore, decided to improve learning analytics dashboards, used in learning management systems, by adding personalized advice based on self-regulated learning strategies, using learning data such as traces of use (e.g. time spent on the environment, connections), social interactions (e.g. number of discussion topics created, number of answers to questions from peers, number of responses to the teacher) collected and stored in the log file. We propose to make a diagnostic to assess whether there is a learning difficulty by observing data related to two types of strategies: cognitive, and task and resource management. A short text will then be generated based on this difficulty and the previous follow-up to encourage the learner in their learning methods. By proposing such advice, we hope to offer a solution to learners while increasing their autonomy and commitment to learning. The paper will present the different processes to be implemented to generate the appropriate advice from the learning data collected in the log files. These processes could be added to the learning analytics dashboard, but also to the learning environment independently the dashboard or still they could be used as a way of direct interaction with learners.

Keywords:

Learning analytics dashboard, personalized advice, self-regulated learning strategies, learning management system.