Complex Networks for Visual Mining of Documents Collections

• Grant number: 09/03306-8
• Support Opportunities: Scholarships in Brazil - Doctorate
• Effective date (Start): August 01, 2009
• Effective date (End): June 30, 2013
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Maria Cristina Ferreira de Oliveira
• Grantee: Robson Carlos da Motta
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil

Abstract

The vast amount of information available in any knowledge domain makes it increasingly difficult to select and analyze specific and relevant information about a specific subject. Visual representations of the collection can be employed to support users in these tasks, and several techniques have been recently proposed to generate visual maps of documents collections. In particular, point placement and multidimensional projection techniques can be used to generate a two-dimensional (possibly three-dimensional) representations of the collection that provide the basis for several visual representations in which the geographical proximity in two-dimensional space indicates similarity of content amongst the documents. Such representations are known as document maps, and provide an appropriate visual interface for the interactive exploration of document collections, featuring a user-driven visual mining process.In this project we propose to investigate how measures defined for complex networks may support visual mining tasks conducted on collections of scientific documents. Networks may be obtained that model different kinds of relationships among the papers. Two hypotheses shall be investigated: (i) whether the integrated use of similarity-based document maps and complex networks obtained from the corpus can decisively contribute in user-driven knowledge discovery and pattern extraction processes, and (ii) whether the analysis of the properties of complex networks that model document similarity can improve the performance of active learning approaches in document classification tasks.

Scientific publications

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

MOTTA, ROBSON; MINGHIM, ROSANE; LOPES, ALNEU DE ANDRADE; OLIVEIRA, MARIA CRISTINA F.. Graph-based measures to assist user assessment of multidimensional projections. Neurocomputing, v. 150, p. 16-pg., 2015-02-20. (09/03306-8, 11/22749-8)

MOTTA, ROBSON; MINGHIM, ROSANE; LOPES, ALNEU DE ANDRADE; OLIVEIRA, MARIA CRISTINA F.. Graph-based measures to assist user assessment of multidimensional projections. Neurocomputing, v. 150, n. B, SI, p. 583-598, FEB 20 2015. (11/22749-8, 09/03306-8)

Academic Publications

(References retrieved automatically from State of São Paulo Research Institutions)

MOTTA, Robson Carlos da. Graph-based measures to assist user assessment of multimensional projections. 2014. Doctoral Thesis - Universidade de São Paulo (USP). Instituto de Ciências Matemáticas e de Computação (ICMC/SB) São Carlos.