Recommending personalized sequences of pedagogical resources to learners: A data mining approach

Location : LORIA lab – Nancy – France (http://www.loria.fr/en/) or LIP6 Sorbonne université

Supervisors :
Armelle Brun, associate professor KIWI team (https://members.loria.fr/ABrun) armelle.brun@loria.fr
Amel Yessad, associate professor MOCAH team, amel.yessad@lip6.fr

Context:
Learners often receive, from their teachers, a set or a list of pedagogical resources to study. These pedagogical resources can be exercises, lectures, quiz, etc.

This set of resources (that can also be an ordered or partially ordered list) is the one that has been defined by the teacher and that fits the best the entire class: profile, past activities, past results, past questions, past difficulties, etc.

Even though this set of resources takes into account the specificities of the class, it does not perfectly fit the profile or the needs of each learner.
Indeed, some learners may not master some basic notions, or may face difficulties with some concepts. At the opposite, they may already master some of them and do not need to study each of the resources.
Similarly, some learners may have some preferred resource types (for example some users prefer studying with exercises instead of lectures), and the set of resources proposed by the teacher may not fit these preferences.

Goal and Scientific challenge:
The goal of this work is to design a personalized path recommendation algorithm, dedicated to learners, where a path is a list (ordered, semi-ordered or non-ordered) of pedagogical resources.

This algorithm relies on the path proposed by the teacher, on the active learner profile, as well as on the profile of all learners.
We propose to adopt a data mining approach, especially to mine frequent patterns.
The algorithm will be dedicated to the mining of multi source and (semi / non)-sequential data, able to evaluate the similarity between profiles, sequences of resources, coping with lack of data or information, etc. and that takes advantage from this possibility to form the personalized path.

Skills
The candidate should have a background in problem modeling, statistics, machine learning and data mining. Programming skills (Java or Python) are mandatory.

PS: Feel free to contact the supervisors for more information.

Lieu: 
LORIA Nancy ou LIP6 Paris
Thématiques: 
Encadrant: 
Armelle Brun
Co-Encadrant: 
Amel Yessad
Référent Universitaire: 
Safia Kedad-Sidhoum
Attribué: 
No
Année: 
2 018

User login