The documents distributed by this server have been provided by the contributing authors as a means to ensure timely dissemination of scholarly and technical work on a non-commercial basis. Copyright and all rights therein are maintained by the authors or by other copyright holders, not withstanding that they have offered their works here electronically. It is understood that all persons copying this information will adhere to the terms and constraints invoked by each author's copyright. These works may not be reposted without the explicit permission of the copyright holder.

Ensuring Novelty and Transparency in Learning Resource-Rcommendation based on Deep Learning Techniques

Author:Wael Alkhatib, Eid Araache, Christoph Rensing, Steffen Schnitzer
Date:August 2018
Kind:In proceedings - use for conference & workshop papers
Book title:in the proceeding of The 13th European Conference on Technology Enhanced Learning.
Keywords:content-based recommender system; word embeddings; deep learning; semantic relations; ontologies.
Research Area(s):Knowledge Media
Abstract:A signi cant di erence in the requirements for recommender systems in the TeL area compared to commercial recommendations lies in the necessity to recommend not primarily similar, but also new items. This is a great challenge for recommender systems. Existing approaches from the family of knowledge-based recommender systems usually use manually created ontologies. In this paper, we present an innovative approach applying deep learning methods. The approach takes into account users' short and long-term interests while ensuring transparency in explaining why a resource is recommended. Our approach relies on Deep Semantic Similarity Model (DSSM) to implicitly measure the semantic similarity between the user interest and the available resources for a recommendation. By taking into consideration the user previous activities, knowledge and current interest, the system reflects the user's history as queries of keywords. In accordance with existing approaches, our process also uses an ontology, but it is created automatically. The experimental results show that our system signi cantly outperforms the baseline in terms of F1-measure and proved its usefulness based on a conducted survey.

If the paper is not available from this page, you might contact the author(s) directly via the "People" section on our KOM Homepage.

[Export this entry to BibTeX]