Abstract:

Most institutions of higher education use limited analytical tools to examine the performance indicators of a student or a class from one semester to the next. Today, using the technological means, students' activities (lessons, simulations, assessment, discussion boards, blogs, etc.) generate thousands of transactions per student that are neither collected or analyzed to locate students who are exhibiting risk to dropping out or failing a course early in the semester. Big Data can provide institutions with prediction tools to anticipate these risks, improve student learning and ensure quality programs.

The objective of our contribution is to set up a technical architecture for a Big Data platform dedicated to the implementation of intelligent teaching.

This platform aims at:
(1) massive recording of textbooks developed by teachers,
(2) monitoring and storing student activities through these textbooks and
(3) using Big Data tools to collect, analyze activities data and
(4) provide reports and statistics to teachers and administrators so that they can
(5) adapt the contents of the textbooks
(6) and follow the students individually in their learning progression.

The modeling solution adopted consists of an hybridization of:
(1) LOM (Learning Object Meta Data) to describe the learning objects forming the textbooks and build a resource catalog to maintain interoperability with other learning management systems and
(2) xAPI (Experience Application Program interface or Tin Can API) to track students' activities to be recorded in an LRS (Learning Record Store) warehouse. Big Data tools collect and analyze textbook content and student activities.

To ensure interoperability with other systems, our platform is based on a Rest API. We have implemented the document-oriented NoSQL database solution as recommended by the LRS standard for storing learning activities. To test our solution, we implemented a real-time monitoring application of the test activities. This application leverages the Streaming processing power guaranteed by our Big Data platform. Our application also allows each teacher to have a dynamic dashboard that displays the results of each student.

Our Application allows the teacher to:
(1) see real-time students with blockages and intervene to help them overcome them.
(2) identify gaps for each student.
(3) and accordingly, to reinforce or adapt the content provided to anticipate any risk of failure for students.

The exploration of the results stored on student activities should allow us in future research to set up software agents for the automatic adaptation of the contents and the real-time monitoring of the students in their learning activities.

Keywords:

Intelligent Teaching, Testing Activity, LOM, xAPI, LRS, Student Classification, Big Data, Streaming Processing, Real-Time Calculation, API Rest, Dashboard.