AbstractThe analysis and visualization of texts have become important subjects in the
face of the enormous production of materials written by the information systems
that exist around us. Everyday, tools and algorithms are developed to help us continue to study and interpret the data we generate.
The research presented here establishes a new metric allowing to compare words from an original corpus. The purpose of this is to have a relevant instrument providing assistance for the terms selection to be confronted during the structural analysis.
At the University of Namur, the EFFaTa-MeM team works on the development
and improvement of a software whose vocation is to help the user in his task of
structural analysis of texts. Although this application - Evoq2016 - offers multiple possibilities, a need related to the proposed possibilities exists. The structural analysis is based on the confrontation of terms, the relevance of the choice of these terms is questioned. Therefore the possibility of assistance for this goal is imagined and it is in this context that the subject of this memoir stands.
The work is an intellectual process based on some fundamental concepts of graph theory accompanied by an implementation in Java programming language. It leads to a motivated inventory and prospects for future work.
In order to provide a way of evaluating the link between the terms of a corpus,
Nexus metrics is developed. It allows embodiment the notion of convergence from one term to another and prepare the spectral analysis computation of the studied graph. It provides an environment for additional researches.
We also justify the relevance of the research area and propose tools for analysis
and data mining within the sphere of structural analysis. Solutions for the visual
representation of the results are highlighted.
|Date of Award||19 Jun 2018|
|Supervisor||Bruno Dumas (Supervisor)|