Побудова рекомендаційних систем на основі онтологій

Loading...
Thumbnail Image
Date
2017
Authors
Жежерун, Олександр
Яремко, Соломія
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
У цій статті наведено опис основних традиційних типів рекомендаційних систем, а також рекомендаційних систем, побудованих на основі онтологій. Проведено огляд способів використання, технологій та засобів реалізації онтологій. Також у статті описано функціонал, структуру та реалізацію рекомендаційної навчальної системи на основі онтологій у вигляді плагіну до системи Protege.
Traditional types of recommender systems – content-based filtering and collaborative filtering – have a number of problems: excessive specialization of domain-based recommendations, a cold start, scalability, data sparsity, and the unusual user problem. Ontology-based recommender systems allow to avoid some of them due to the semantic reasoning – capabilities that are provided by the description of logic reasoners that conduct a semantic analysis of ontologies in order to improve the quality and truthfulness of recommendations. This is a special type of recommender systems based on the clear knowledge of the elements intended for recommendations, user needs, and recommendation criteria. The formal basis of the OWL – ontologies description language – are description logics that provide the possibility of logical reasoning. Taking into consideration the fact that the Protégé ontology editor uses the OSGI framework as a plugin infrastructure, which allows programmers to integrate plugins that can be displayed as separate tabs, it was decided to implement the system of giving student recommendations on choosing of selective disciplines of the NaUKMA informatics department on the basis of explicitly prescribed student restrictions in the form of WorkspaceTab – Plug-in Protégé 5.1.0 using the Java programming language. Since the search for recommendations should be based on the clearly defined user constraints, the system is implemented as a constraint-based recommender system based on ontologies. The functionality, structure, implementation, the main problems that were encountered during the system implementation, and the prospects for its further development are identified in the article.
Description
Keywords
рекомендаційна система, онтологія, дескриптивна логіка, плагін, recommendations, ontology, descriptive logic, ontology editor, Protégé, plugin, стаття
Citation
Жежерун О. П. Побудова рекомендаційних систем на основі онтологій / Жежерун О. П., Яремко С. А. // Наукові записки НаУКМА. Комп'ютерні науки. - 2017. - Т. 198. - С. 36-41.