Développement d’un système de recommandation à base de connaissances pour le domaine de Tourisme

Abstract

Our graduation project was to design and build a knowledge-based recommender system for the Tourism domain, by using the “DATAtourisme” ontology as a new and rich knowledgebase in this domain. Our recommender system can work on mobile, web and desktop platforms. Thanks to the gained benefits by using the Google project “Flutter”. To carry out this work, we had to go through several phases. At first, we did research on the field of study, this research focused on generalities of recommender systems and the basic designed models for them, such as Collaboratif filtering and Content-based models. Then we did a research on the field of knowledge-based recommender systems, we explained what is knowledge, knowledge-based, and we clarified the main forms to represent knowledge. Moreover, we gave a brief explanation of the two types of knowledge-based recommender systems. Concerning the design phase, we gave a full description of DATAtourisme ontology, and constraint-based recommendation as an adopted approach. This phase ended by highlighting our architectural design for the proposed recommender system with appropriate descriptions for its modules inclosing several formal algorithms. In implementation phase, we presented our development environment by indicating various tools and technologies which are used, such as Flutter, API DATAtourisme, firebase, etc. then we proved the execution process of our application by depicting the obtained results while presenting some screenshots (interfaces) according to several uses scenarios. Finally, we cited some advantages and limits of this research work. As far as we can tell is that the Knowledge-based recommender systems requires a big efforts in term of system design, they use a lot of techniques (not simple as in content-based or collaborative-based filtering). In addition, Knowledge-based recommender systems help users to explore and understand the domain knowledge. In this kind of recommender system, users are an integral part of recommendation process knowledge, by developing their information needs during their frequent interactions with the recommender system. This project was a good opportunity to discover and deepen our knowledge domain, recommender systems, and to push our skills by using further new technologies.