Visualização em nuvens de texto como apoio à busca exploratória na web

• Autor(a) principal: Lunardi, Marcia Severo
• Data de Publicação: 2008
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da UERJ
• Texto Completo: http://www.bdtd.uerj.br/handle/1/9166
• topic: Text Clouds; Information Visualization; Information Design; Interface Design; Web Search; Exploratory Search; Information Retrieval; Nuvens de Texto; Visualização de Informações; Design da Informação; Design de Interface; Sistemas de Busca; Busca Exploratória; Recuperação de Informações; CNPQ::CIENCIAS SOCIAIS APLICADAS::DESENHO INDUSTRIAL

This dissertation presents the results of a research that evaluates the advantages of text clouds to the visualization of web search results. A text cloud is a visualization technique for texts and textual data in general. Its main purpose is to enhance comprehension of a large body of text by summarizing it automatically and is generally applied for managing information overload. While continual improvements in search technology have made it possible to quickly find relevant information on the web, few search engines do anything to organize or to summarize the contents of such responses beyond ranking the items in a list. In exploratory searches, users may be forced to scroll through many pages to identify the information they seek and are generally not provided with any way to visualize the totality of the results returned. This research is divided in two parts. Part one describes the development of an application that generates text clouds for the summarization of search results from the standard result list provided by the Yahoo search engine. The second part describes the evaluation of this application. Adapted to this specific context, a text cloud is generated from the text of the first sites returned by the search engine according to its relevance algorithms. The benefit of this application is that it enables users to obtain a visual overview of the main results at once. From this overview the users can obtain keywords to navigate to potential relevant subjects that otherwise would be hidden deep down in the response list. Also, users can realize by visualizing the results in context that his initial query term was not the best choice.