Similaridade Semântica
Rita Carolina Costa, Thiago Henrique Bragato Barros, Renato Fileto · Brazilian Journal of Information Science research trends · 2024
In the rapidly evolving field of Natural Language Processing (NLP), understanding the domain of semantic similarity is of paramount importance for both academic and industrial applications. This article presents a comprehensive domain analysis of semantic similarity, integrating a multidisciplinary approach that encompasses key concepts, interrelations among these facets, stakeholders, information practices, and existing classification systems. We elucidate the core ideas, such as lexical and syntactic similarity, embeddings, and various similarity metrics, and demonstrate their interrelatedness. The paper also identifies and characterizes the diverse array of stakeholders involved in this domain, from academic researchers and tech leads to policymakers and open-source communities. Furthermore, we explore how information is disseminated and used within this domain, including an examination of research publication trends and industry reports. Lastly, the article assesses existing classification systems and ontologies that structure the knowledge in this field. Our findings serve as a foundational framework for future research, development, and ethical considerations in the semantic similarity domain. This in-depth analysis aspires to guide both newcomers and seasoned experts through the intricate landscape of semantic similarity, thereby contributing to the field's holistic advancement.