Information architecture applied on natural language processing: a proposal: Information Science contributions on data pre-processing for training and learning of artificial neural networks

• Autor(a) principal: Kuroki Júnior, George Hideyuki
• Data de Publicação: 2023
• Outros Autores: Duque, Cláudio Gottschalg
• Tipo de documento: Artigo
• Idioma: por; eng
• Título da fonte: Revista Brasileira de Direito Civil (Online)
• Texto Completo: https://periodicos.sbu.unicamp.br/ojs/index.php/rdbci/article/view/8671396
• topic: Ciência da Informação; Arquitetura da informação; Tratamento da informação; Inteligência Artificial; Processamento de linguagem natural; Information Science.; Information architecture; Information treatment; Artificial Intelligence; Natural language processing

Introduction: Natural Language Processing with artificial neural networks has gaps that can be addressed by Information Science. Objective: To propose Information Science contributions through Multimodal Information Architecture applied to information organizing focused on artificial neural networks training, positioning it as an active body of knowledge in artificial intelligence problems. Methodology: Adapting a path of three levels of analysis (metaphysical, scientific and technological), verify the current stage of development of natural language processing techniques (metaphysical); use definitions from Multimodal Information Architecture proposing a five-step procedure for designing, analyzing and preparing information spaces for training and learning methods of neural networks, fulfilling gaps identified by authors focused on Computer Science implementations scientific); verify proposal applicability through 3 sets of data from 16 knowledge areas as a basis for evaluation (technological). Results: Results obtained through pre-processed data and raw data where compared. Great potential in developing a structured method of Multimodal Information Architecture was observed, which would provide instruments to organize data which are used as test and learning sample in artificial neural networks. Conclusion: This method could place Information Science as an actor and producer of data pre-processing solutions, replacing its current role as consumer of prefabricated solutions made by Computer Science.