Saroglou Stylianos (Phd Candidate)

Thesis title: Entity extraction and multi-label classification with machine learning methods

Supervisor: Goulianas Konstantinos

Advisory Committee Members:

Konstantinos Diamantaras, Professor, Dept. of Information and Electronic Engineering, IHU
Bratsas Charalampos, Ass. Professor, Dept. of Information and Electronic Engineering, IHU

Abstract:

The study of this thesis focuses on the extraction of entities from written texts and their classification into relevant knowledge bases (ontology). By using modern machine and deep learning algorithms in plain text, the entities referred to it are automatically recognized and then their meaning is clarified, making the correspondence in the ontology. The purpose of this research is the analytical mathematical documentation of the various algorithms, techniques and data for dealing with entity recognition and entity linking, but also of multi-label classification, i.e. problems in the field of information extraction from texts. Specifically, the applicability in multiple domains is explored, such as the labor market, legal texts and more general knowledge bases (WikiPedia). It becomes one comparative evaluation aimed at the construction of a holistic model, capable of dealing with the above problems effectively regardless of the application domain.