Terminologia e estratégias de busca : um estudo sobre a utilização da linguagem natural e de termos controlados nas pesquisas do Portal de Periódicos da CAPES

Abstract

The objective of the research was to present a proposal for optimizing searches on the CAPES Portal using natural language and controlled terms, aiming to develop skills in creating information search strategies. Natural language, used in everyday contexts, follows rules inferred from habitual use. Controlled terms, or terminological language, represent specialized knowledge, offering conceptual precision in professional communication and greater efficiency in information retrieval. Although academic training equips librarians to create search strategies across various disciplines, these skills are rarely applied in practice. User training faces time and frequency constraints, hindering the broader academic community's ability to adopt bibliographic and terminological research practices. The research was applied, descriptive, and employed a case study approach. Data collection occurred in three phases: extracting keywords from the abstracts of dissertations from the Graduate Program in Language Studies; converting these keywords into technical terms using terminological dictionaries, glossaries, and thesauri; and conducting searches on the CAPES Portal, comparing results obtained with natural language keywords and specialized terms. The methodology combined direct data collection and exploratory research, with a mixed analysis approach. Effectiveness was evaluated through precision calculations, defined as the ratio of relevant documents retrieved to the total documents retrieved. Relevance was determined based on the objectives outlined in the abstracts. Results were presented in tables for easier interpretation. Controlled terms were sourced from tools such as the Thesaurus Brasileiro da Educação (BRASED), Dictionnaire de Linguistique, UNESCO Thesaurus, and Tesauro ROSSIO, all within the linguistics field. Thirty-four searches were conducted using natural language keywords, and another 32 with controlled terms. Controlled terms demonstrated greater effectiveness, retrieving relevant results in 40.62% of searches, compared to only 12.5% for natural language. In 46.88% of searches, no relevant documents were retrieved, as they pertained to areas unrelated to linguistics, such as medicine, nursing, and business administration. Even content in the education field did not fully meet the proposed objectives, requiring complementary strategies to filter information. The study concluded that controlled terms enhance information retrieval efficiency, although their superiority is not absolute. In certain situations, natural language proved more effective, particularly in the initial stages of data collection. Combining natural language keywords with technical terms, identified through terminological dictionaries, proved essential for refining search strategies, facilitating 11 information retrieval, and contributing to the creation of more precise texts. Moreover, the use of Boolean operators ("AND," "OR," and "NOT") and advanced filters refined searches, balancing the number of retrieved documents and reducing the effort required to analyze irrelevant information. The research demonstrated that integrating different search approaches and tools significantly improves the efficiency of academic information retrieval.