Estudo e aplicação de metodologia para reengenharia de tesauro: remodelagem do THESAGRO

• Ano de defesa: 2015
• Autor(a) principal: Benildes Coura Moreira dos Santos Maculan
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Tesauro; Modelo de conversão de tesauro; Reengenharia de tesauro; Sistema de organização do conhecimento; Modelagem conceitual; Recuperação da informação; Tesauros; Sistemas de organização; Ciência da Informaçao; Organização da informação; Reengenharia de sistemas
• Link de acesso: http://hdl.handle.net/1843/BUBD-9ZKMUV

Knowledge organization systems (KOSs) encompass different types of representation languages such as taxonomies, thesauri, and ontologies. These instruments have different levels of terminological control and indicators of relationships between concepts. These relationships may be used to help data integration in a web environment. Our work focuses on the thesaurus, which is defined by a controlled and specialized vocabulary, consisting of a set of descriptors that are interrelated by equivalence, hierarchy, and association relationships. We investigate and apply a model for reengineering a traditional thesaurus, with the goal of transforming the concept relationship system of the thesaurus into a semantically richer structure. Our main contribution is to demonstrate the viability of this conversion as an aid to information organization and management. To provide the theoretical context of our work, we review the thesaurus construction literature, and discuss several approaches and their different meanings. We provide an overview of the Library and Information Science (LIS) field, covering the foundations of Terminology, Semantics, and Pragmatics, as well as the foundations of Qualia roles and the Valence Theory, which was developed by Borba (1996), and assigns semantic values to Portuguese verbs. In this overview, we discuss several themes, including concept, term, definition, ambiguity, and polysemy. Our research environment is the Brazilian Agricultural Research Corporation (EMBRAPA), which is responsible for developing technological solutions for information sharing in the Brazilian agriculture domain. Our methodology consists of a case study and involves exploratory, applied, and empirical procedures combined with a qualitative analysis of the specialized literature of the agricultural and the thesaurus construction fields. The thesaurus reengineering model we apply was developed by Soergel et al. (2004) and Lauser et al. (2006), and consists of procedures that allow us to process the different entities (concepts, terms, strings, scope notes, and relationships) of the conceptual structure of a traditional thesaurus through a formal refinement of the relationships between concepts. We describe and analyze in detail these procedures of the reengineering model. We apply the reengineering model to THESAGRO, which is a Brazilian thesaurus in the agriculture domain, in the thematic frame of agriculture intensification, using taxonomy as a basis and other agriculture thesauri as inputs. In the process of applying the model, we use the e- Termos system, which is a software application that helps with the development and management of terminological products. The results demonstrate the viability of using the model to reengineer a traditional thesaurus into a semantically richer structure that specifies the types of relationships between concepts and terms. The results also show that the construction of the structure of thesauri should be done in a more formal way, based on international conventions and represented in conceptual modeling languages such as Simple Knowledge Organization System (SKOS) and Resource Description Framework Schema (RDFS), to promote interoperability among different thesauri. Finally, the ability of refining the relationships between concepts, by explicitly modeling them in the thesaurus structure, helps us organize the knowledge in the modeled domain, which in turn helps users explore the domain, because it makes these relationships clearer and more specific.