MSCS Thesis

Date: 15 July 2008
Master’s Degree in Advanced Computer Systems and Network Services
Score: 110/110 cum Laude

Thesis title: “Enhancement of the Case-Based Reasoning framework jCOLIBRI’ with a Pivoting-Based Retrieval algorithm and Time Series features”.

Description: The thesis work (6 months) consisted in the extension of jCOLIBRI ( http://gaia.fdi.ucm.es/research/colibri/jcolibri ), an open source object-oriented framework for the realization of Case-Based Reasoning (CBR) systems, developed by GAIA (Group for Artificial Intelligence Applications ) of the University Complutense of Madrid.
– Development of a Pivoting-Based Retrieval algorithm, which enhances the recovery of similar cases, reducing the number of comparisons with the new input case.
– Pivoting-Based Retrieval algorithm performance tests and results analysis.
– Development of a new type of attribute represented by Time Series and functions implementation for the calculation of the similarity on this type of attributes.
– Integration of functions for Time Series dimensionality reduction, such as Fourier transformation and Piecewise transformation, in order to make easier the comparison between this type of attributes.
– Testing of the developed capabilities on a real dataset, representative of dialysis sessions data, provided by Pavia hospital.

Key technologies and tools: JAVA, Hibernate, PostgresSQL, Eclipse IDE, jCOLIBRì framework, HSQLDB (RDBMS used for testing)