Protótipo de rede social focada em debates: ArgViewer

• Autor(a) principal: Silva, Denis Castro da
• Data de Publicação: 2022
• Outros Autores: Santos, Guilherme Nunes Amaral dos, Francesconi, João Carlos Palacios, Vasconcelos, Vinicius Prado
• Tipo de documento: Trabalho de conclusão de curso
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da PUC_SP
• Texto Completo: https://repositorio.pucsp.br/jspui/handle/handle/42575
• topic: CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO; Argumentação suportada por computador (CSA); Processamento de linguagem natural (NLP); Similaridade de sentenças; Inteligência coletiva; argumentação; Social network; Computer supported argumentation (CSA); Natural language processing (NLP); Sentence similarity; Collective intelligence; argumentation

The number of people connected to the Internet is increasing, as is the number of interactions between them. But virtual (and even face-to-face) discussions and debates are still not very effective and/or efficient. Dialectics is one of the oldest and most effective ways to generate reflection and knowledge, but current technologies and social networks have only increased the quantity of dialogue, leaving quality aside. Therefore, this work sought to create a prototype of a social network for debates, trying to use the available technologies to optimize the debate process. The methodology of the work is a case study, and the software development process followed the 5 basic activities of software engineering: communication, planning, modeling, construction and deployment. In the modeling stage, the analysis was done focusing on use cases and the software design followed the domain-driven design (DDD). The system created seeks to organize the debate clearly and simply, through brief propositions that relate as pros or cons to each other. A crowdsourcing voting system is used to evaluate the veracity and relevance of the propositions. In addition, a semantically similar propositions filter (implemented by a machine learning transformer model) is used to decrease the number of repeated contributions. And an offense filter with SVM model is used to prevent the creation of inappropriate contributions. The application was implemented with Java Spring Boot and Python on the backend and React-Bootstrap on the frontend. Finished, the prototype created shows some possible strategies to try to create a social network of rational debate