Análise de assuntos assistida por computador: proposta metodológica de um modelo de inteligência aumentada

Abstract

Theme: Augmented Intelligence is the expansion of human cognitive faculties resulting from the use of intelligent tools. The origin of the concept is attributed to Douglas Engelbart who, in 1962, proposed a methodological framework on how to increase human intellect to solve complex problems with the assistance of tools. Object: The researched object was Computer Aided Subject Analysis. Scope: In the context of Information Organization, the application of Augmented Intelligence seeks to solve the dilemma between adopting the speed of automation or maintaining the quality of intellectual indexing when processing the high volume of textual data available. Objective: To develop a methodological proposal for an augmented intelligence model to enable computer-assisted subject analysis. Setting: The collection of the Agricultural Research Database (BDPA), having as a cut a set of 7377 scientific articles in agricultural research indexed by the Thesaurus Agricola Nacional (Thesagro). Research Type: This research is applied in nature; as for the objectives, it is exploratory research; and as for the methodological procedures, it is experimental research. Data Collection: The data collected were scientific articles and their descriptors attributed by people from the Brazilian Agricultural Research Corporation (Embrapa) and published between 1998 and 2021. All data are available for public access and free of charge. Methods: Artificial Intelligence methods of Natural Language Processing were used to perform automatic summarization, identification of text subjects and topic modeling to produce Computer Aided Subject Analysis. Theoretical Background: Artificial Intelligence can produce good technical results based on data and Human Intelligence can use these results as subsidies for decision making in the process of indexing scientific articles subjects. Historical/Contextual Background: Scientific literature on Augmented Intelligence was collected and selected from the scientific databases Library and Information Science Abstracts (LISA); Library, Information Science and Technology Abstracts (LISTA); Institute of Electrical and Electronics Engineers (IEEE); Scopus; Web of Science. Author's Point of View: The point of view on Augmented Intelligence for Computer- Aided Analysis of Subjects belongs to the author of the research, as it is an unprecedented approach, an innovation in the area. However, the concept of Augmented Intelligence is from Bush (1945) and Engelbart (1962). Cause and Effect: The use of Artificial Intelligence to extract insight into the indexing data from a database formulated by an expert allowed producing a subject analysis with satisfactory results of over 79%, which associated with Augmented Intelligence can reach 100%. Results: A technological prototype and a Database were developed, presented under a unified theoretical-methodological proposal to technologically support decision-making in the Translation stage of the indexing process. Thus, Augmented Intelligence, as a design pattern, provided the theoretical constructs for the implementation of the technological prototype called Computer Aided Subject Analysis (3ApC), which was written in open-source code in the Python programming language. The validation experiment of the developed model used 7073 scientific articles and their respective indexing terms from the Agricultural Research Database (BDPA). It was concluded that the use of 3ApC adequately assisted the Analysis of Subjects and could speed up the indexing process with the same quality as the intellectual process. In addition, it broadened the horizon of knowledge in Information Science and added technological innovation to the set of tools available for the Organization of Information.