A recommender system to support the development of context-aware intelligent transportation systems

• Ano de defesa: 2022
• Autor(a) principal: CHAGAS, Amirton Bezerra
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal de Pernambuco; UFPE; Brasil; Programa de Pos Graduacao em Ciencia da Computacao
• Palavras-chave em Português: Redes de computadores e Sistemas distribuídos; Sistemas de recomendação; Sensibilidade a contexto; Sistemas de transporte inteligente; Taxonomia
• Link de acesso: https://repositorio.ufpe.br/handle/123456789/46362

The development of Context-Aware Intelligent Transportation Systems (ITS) requires a careful analysis to identify which Contextual Elements can contribute to the definition of the application’s Context. This activity is complex, especially in the ITS scenario, very vast and with hundreds of possibilities. The objective of this research is to analyze the use of context-awareness in ITS and propose alternatives for organizing this information to allow the creation of tools that contribute to automating part of the task of identifying useful contextual elements for the development of a new application. A literature review of ITS projects served to map the use of contextual elements. 73 projects were mapped, of which 70 were academic and 3 were commercial. Then, a Taxonomy of Contextual Elements Categories was defined, to increase the granularity of the information and facilitate its use in an automated system. The taxonomy has 79 categories in total. A knowledge base was built relating the 73 projects to the taxonomy categories. An experiment showed an increase of 197.5% in the amount of contextual elements correctly chosen when designing an application, when the engineer has knowledge and access to the taxonomy. Using the taxonomy and the knowledge base, we designed a Contextual Element Category Recommender System for ITS. Using an initial subset of Contextual Elements already identified as necessary for a new application, it can recommend categories of Contextual Elements for the subsequent analysis by the application designer. The recommender system validation demonstrated its ability to recommend categories relevant to projects. When using a number n >= 8 of similar projects to identify the categories, even limiting the number of recommendations to 15 items, in more than 75% of the time the system recommended categories known to be used for the subset informed as input. The creation of a taxonomy associated with the development of a recommender system using a knowledge base of projects in the ITS area presented the potential to contribute positively to the design and development of applications in this domain, allowing the identification and consequent use of more relevant contextual elements for the project application.