Skip to main content

Header

Date
05.10.2022 | 14:00 - 15:00 (CET)

AI-Cafe presents: Named Entity Recognition: Using deep-learning approaches for extracting entities

Andel Gugu
(Researcher at Fraunhofer IAIS, Germany)

NER Taxonomy

We will briefly introduce the deep-learning approaches that tackle the task of NER, followed by a comparative evaluation between multiple setups of a chosen network architecture. The biggest available datasets for German, CoNLL-2003 and GermEval2014 will be used to train and evaluate the NER models. Finally, we will see how transfer learning can be applied, to improve the performance of pre-trained models in "low-data" settings.

This technology is listed in the AI Assets Catalog.

Speakers

Andel Gugu

Speaker's short bio:

My name is Andel Gugu, and I work as a research engineer. I have a Master’s Degree in Computer Science from the University of Bonn, where the main track of my studies was ‘Intelligent Systems’. During my studies, I specialized in courses covering machine learning, artificial intelligence, and neural networks. My interest in these subjects aligned with that of Fraunhofer IAIS, where I started working as a student assistant in November 2019. During my work, I became familiar with different media analysis services and the technologies behind them. I wrote my thesis on ‘DL-based Entity Recognition models’ under the supervision of ‘The Media Analysis Platform’ team’s leader. Last year, I switched from being a student assistant to working as a full-time employee. I am now part of the team that develops and maintains the Mining Platform, but I also work on improving the quality of some of the existing media services provided by this platform.