Mineração de soluções relevantes para tarefas de programação a partir de resultados de mecanismos de busca

• Autor(a) principal: Rocha, Adriano Mendonça
• Data de Publicação: 2022
• Tipo de documento: Tese
• Idioma: eng
• Título da fonte: Repositório Institucional da UFU
• Texto Completo: https://repositorio.ufu.br/handle/123456789/36237; https://doi.org/10.14393/ufu.te.2022.565
• topic: Tarefas de Programação; Programming Tasks; Mineração de Soluções Relevantes; Mining Relevant Solutions; Motores de Busca; Search Engines; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO; Computação; Engenharia de sistemas; Ferramentas de busca na Web; Programadores de computador

Software development is a knowledge-intensive activity. Official documentation for developers may not be sufficient for all developer needs. Searching for information on the web is a usual practice, but finding really useful information may be challenging, because the best solutions are not always among the first ranked pages. So, developers have to read and discard irrelevant pages, that is, pages that do not have code examples or that have content not focused on the desired solution. This work aims to understand how the ranking quality returned by search engines can influence the performance of developers when solving programming tasks, and to propose an approach to mine relevant solutions for programming tasks from search engines results. In a preliminary analysis, we evaluated the top 20 pages returned by Google’s search engine, for 10 different queries, and observed that only 31% of the pages evaluated are relevant to developers. Therefore, we carried out a first study with developers that showed that they spent less time solving programming tasks, when using a higher quality ranking (on average, developers spent around 4 minutes more during solving tasks, when using a lower quality ranking). In a second study, we proposed and evaluated three different approaches to mine relevant pages returned by the search engine. The Google search engine that was used as a baseline returns a fair number of pages that are irrelevant to developers. The filters we propose are effective in removing irrelevant pages, in this way, a better quality ranking is generated. Developers can benefit from these filters in order to increase their productivity while performing programming tasks.