Banca de QUALIFICAÇÃO: VICTOR ALEXANDRE PLOEGER MANSUELI
Uma banca de QUALIFICAÇÃO de DOUTORADO foi cadastrada pelo programa.STUDENT : VICTOR ALEXANDRE PLOEGER MANSUELI
DATE: 13/09/2022
TIME: 14:00
LOCAL: Remoto, via GoogleMeet https://meet.google.com/ckq-uymh-kbw
TITLE:
Graph-based anomalies in the formation of academic committees: detection and characterization
PAGES: 72
BIG AREA: Ciências Exatas e da Terra
AREA: Ciência da Computação
SUBÁREA: Metodologia e Técnicas da Computação
SUMMARY:
The degree of masters and doctors in the world is a facet of the formation of human resources, and intellectual production in the form of dissertations and theses is one of the essential and indispensable components for the permanent evolution of Science. Academic committees, as evaluation instances with power of rejection, are responsible both for the proficiency of the human resource that is being formed, as for the contribution that its research can offer to the social environment in which it is inserted. Due to the dynamic nature of the composition of these committees, we propose to analyze them empirically, providing a better understanding of their role as an integral agent of scientific knowledge. Making use of graphs in committees modeling, we can treat the relationships between the participating members as a social network, according to the role played by them (evaluatees, evaluators or advisors), benefiting from existing methods, techniques and algorithms. Four models of academic graphs (a-graphs) were formalized to represent these relationships: (i) evaluations, (ii) invitations, (iii) co-participations and (iv) orientations. These different configurations in the committees composition can generate recurring (sub)structures (patterns) and deviations from what is considered regular (anomalies). Detecting and characterizing these patterns and anomalies are the main objectives of this research project. For this, we developed a Knowledge Discovery Process, systematizing the necessary steps to achieve this knowledge: data collection; pre-processing; transformation; graph mining; visualization and analysis. Two real-world datasets were used for experimentation: the Catalog of Theses and Dissertations from CAPES, with more than half a million manuscripts collected; and the National Repository of French Electronic Theses (STAR). In a theoretical-conceptual perspective, we also proposed a typology to qualify anomalies in a-graphs according to their levels, types and subtypes. Our preliminary results using the Graph-Based Anomaly Detection (GBAD) System were promising, with anomalies detected by the algorithms that address the MDL (Minimum Description Length) selection principle, both structural and categorical ones. With this research project we hope to contribute to deepen the understanding of the dynamics involved in the composition of academic committees, not only in Brazil but also abroad.
BANKING MEMBERS:
Presidente - Interno ao Programa - 1934625 - JESUS PASCUAL MENA CHALCO
Membro Titular - Examinador(a) Externo à Instituição - ROBERTO HIRATA JUNIOR - USP
Membro Titular - Examinador(a) Externo à Instituição - FABIO CASTRO GOUVEIA - FIOCRUZ - RJ
Membro Suplente - Examinador(a) Externo à Instituição - LUCIANO ANTONIO DIGIAMPIETRI - USP