Approaches for the automatic linking of web repositories

Author(s)

Jose Antonio Camacho Guerrero

• Document type: Master's Dissertation
• Press: São Carlos. , gráficos, ilustrações, tabelas.
• Institution: Universidade de São Paulo (USP). Instituto de Ciências Matemáticas e de Computação (ICMC/SB)
• Defense date: 2002-12-20

Examining board members

Maria da Graça Campos Pimentel; Wagner Meira Junior; Maria das Graças Volpe Nunes

• Advisor: Maria da Graça Campos Pimentel
• Field of knowledge: Physical Sciences and Mathematics - Computer Science
• Indexed in: Banco de Dados Bibliográficos da USP-DEDALUS
• Location: Universidade de São Paulo. Instituto de Ciências Matemáticas e de Computação. Biblioteca Prof. Achille Bassi; ICMSC/T; G934ia

Abstract

Many everyday experiences aro based on lhe exchange of information which can be captured to support the automatic generation of multimédia hyperdocument.s corresponding to those experientes. The eClass Project, developed at. the Geórgia Inslitute of Technology, exploit s ubiquitous computing technologies to capture information in classrooms towards allowing the automatic generation of hypermedia documents corresponding to live lectures. Such environments allow the automatic generation of hypertext. links from the interaction between users and their ubiquitous computing environments. When the documents generated are made available in the Web, there is a, demand for services that support. the automatic indentification of hypertext links among the documents themselves or among those documents and other web-based material; such services can make use of Information Retrieval techniques. However. Information Retrieval techniques usually lead to the identification of relationships that should not have been generated (generating non-relevant links) at the same time that fail to identify ali relevant relationships (poor recall). The aim of this work is to build upon automatic linking services in order to further investigate information retrieval techniques towards improving the quality of the relationships identified. The investigations have used a collection of documents from online Brazilian Newspapers and t.he Cvstic Fibrosis Collection. The relationships identified have been used to generate links among t.he repositories investigated, while the links have been stored in an open hypermedia. linkbase to allow their use by any applications. The positive results of the investigations have defined the requirenients for a configurable automatic linking service made available as a contribuition of this work. (AU)