Serendipity prospecção semântica de dados qualitativos em Educação Especial

• Ano de defesa: 2016
• Autor(a) principal: Fernandes, Woquiton Lima
• Orientador(a): Mendes, Enicéia Gonçalves
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos; Câmpus São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Educação Especial - PPGEEs
• Palavras-chave em Português: Educação especial; Análise qualitativa; Ontologia computacional; Mineração de textos; Anotação semântica
• Palavras-chave em Inglês: Special education; Qualitative analysis; Computational ontology; Text mining; Semantic annotation
• Área do conhecimento CNPq: CIENCIAS HUMANAS::EDUCACAO
• Link de acesso: https://repositorio.ufscar.br/handle/ufscar/8576

In the past decades, there has been a revolution in the way science has been conducted. The current context has demanded more collaborative work such as, studies in research networks of large scale. One of the many essential marks of change in this new way of making science has been the intense usage of Information and Communication Technologies (ICT), or “eScience”. Nowadays, it plays a fundamental role in the methodology adopted by many research groups around the world. Analyses of the qualitative data evidenced in researches about Special Education were done then. The biggest challenge that was noticed would be to advance in the analysis of qualitative data using information technologies without losing the subjectivity involved in the research and to broaden the capability of going over the data without losing the right to come and go, the right to critique and establish proper reflexions, respecting subjective positioning and, above all, maintaining the research's critic criteria. In this sense, this work establishes as its main objective to evaluate the proposed technological architecture of qualitative analyses of data. This analysis was based upon data mining theories, researches in ontology and techniques of semantic notation in the field of special education aiming to analyze the thresholds and possibilities this methodological approach permits. We used as methodology the construction of a prototype, named Serendipity, based on the perspective of software engineering, in order to extract the main techniques that could set as a safe method for design, implementation and deployment of the solution. Cyclically, the methodology allowed us to modify requirements and establish improvements, allowing the feedback process from new analyses. The text mining process relied on gaining knowledge from textual databases that have little or no data structure. The computational ontology was the element able to reconstruct the syntactic representation, giving it direction. The words (data) are related and are set within a context of formal knowledge, providing them with a semantic and cognitive ability, building concepts, open to interpretation, comprehension and common understanding; as a result, we built up a specific ontology for Special Education. The semantic annotation helped attach content to the text to describe their semantics, allowing that software agents could retrieve information in a more precise manner through the association of the document to the ontology in a conception of semantic fields. We built a customized dictionary for special education to relate terms to synonyms and expressions associated with the ontology. To view beyond the semantic classes, we used automatic concept maps to establish relationships between concepts included in a hierarchical structure of propositions. Finally, to assess the proposal, we made use of part of the data collected from the National Observatory of Special Education in transcribed texts about the formation of five cities, one from each region of Brazil. The results show limits already recognized in the proposal and; in this respect, did not aim to establish a subjective and deep analysis that would permit extreme precision results. It points out that the researcher is and will always be the driving factor that operates the process’ flow and relying, or not, on computing tools is not entirely immune to err. The proposal of serendipity has given a step forward in the automatic process of data analysis and can be used in big data without losing the subjectivity of the researcher. However, we must add new human and technological resources to contribute to its improvement and encourage other areas to develop domain ontologies with their experts and the development of specific dictionaries. Therefore, despite its limitations, the approach has shown significant advances in semantic exploration of qualitative data in the Special Education field and it is capable of being adapted to other areas and fields of knowledge.