COMPARED THEMATIC REPRESENTATION IN DIGITAL GOVERNMENT

Abstract

The digital transformation of public services amplifies the challenges of organizing and retrieving information, especially given the emergence of new concepts and the diversity of information objects. This study comparatively analyzed the subject representation of textual and non-textual information objects in the Digital Government domain, considering the contributions of natural language, controlled vocabulary, and artificial intelligence regarding semantic adherence, terminological specificity, and the occurrence of inadequate subject inferences. The research, applied and exploratory in nature, adopted a qualitative and quantitative approach and was structured as a comparative subject representation experiment. The corpus comprised 16 information objects, 8 textual and 8 non-textual, from which 174 term occurrences were analyzed. The procedures involved identifying subjects in natural language, terminological and semantic validation supported by specialized glossaries, comparison with a controlled vocabulary, and analysis of terms generated by an artificial intelligence tool. The results indicated greater adherence between natural language and specialized glossaries than between natural language and the controlled vocabulary, limitations of the latter in representing emerging concepts, and the potential of artificial intelligence as a support resource, although divergences and unverified inferences were identified. It is concluded that, in the Digital Government domain, the articulation between natural language, controlled vocabulary, and artificial intelligence, combined with terminological updating and human validation, can contribute to a more consistent subject representation.