Alteração - Atendimento do Repositório Institucionalclique aqui para acessar a portaria
Please use this identifier to cite or link to this item:
|metadata.eventos.dc.title:||Automatically generated summaries as in-page web navigation accelerators for blind users|
|metadata.eventos.dc.creator:||Alves, Erick Antonio|
Cardoso, Paula Christina Figueira
Freire, André Pimenta
Automatic text summarization
|metadata.eventos.dc.identifier.citation:||ALVES, E. A.; CARDOSO, P. C. F.; FREIRE, A. P. Automatically generated summaries as in-page web navigation accelerators for blind users. In: BRAZILIAN SYMPOSIUM ON HUMAN FACTORS IN COMPUTING SYSTEMS, 17., 2018, Belém. Proceedings… New York: ACM, 2018. Não paginado.|
|metadata.eventos.dc.description.abstract:||Blind users often experience a high information overload when browsing through textual content on long Web pages. This occurs mainly because these users browse the Web using screen reader software that reads line by line text content sequentially, causing the user to listen to much of the content until finding what is relevant, especially when there are scant structural elements (such as headings) to help them. In order to reduce the information overload of blind users on the Web, the aim of this work was to develop and evaluate an application that provides these users with an automatic summary with the main information of the page, allowing a direct navigation between the summary and the text through anchors to enable faster navigation. The paper explored three different summarization approaches for Brazilian Portuguese. To evaluate navigation aspects and the most appropriate types of summaries, user tests were performed with three blind users and four normalvision users. The results showed good acceptance of the prototype, with indication of the preferred summarization approaches and implications for design to help users better use those resources in tasks where skimming long texts is necessary.|
|Appears in Collections:||DCC - Trabalhos apresentados em eventos|
Files in This Item:
There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.