Arbeitsgruppe Informationsmanagement
31.05.2021

Machine Learning For People With Reading and Writing Difficulties

A large number of people struggle with reading and writing. A 2011 study by the University of Hamburg found that 4% of German adults are totally illiterate and more than 14% are functionally illiterate, i.e. they are unable to use writing in everyday life in a way that is expected in the social context. Together with the University of Bremen and the University of Bayreuth, we are working on HCI+AI solutions to support people with reading and writing difficulties.

In this thesis project, you have the opportunity to develop and evaluate an interactive system that automatically summarizes text for this use case. There is the opportunity to build on and extend an existing prototype of such a system. The system provides users with the most important information contained in the text. Based on your individual interests, you can either work on 1. applying and extending state-of-the-art machine learning (ML) systems like BERT, PEGASUS, and GPT-2, or you can work on 2. designing and evaluating user interfaces for such ML systems in this context.

Requirements:

Experience with machine learning (especially neural networks) and user-centred design.

 

Contact:

Dr. Hendrik Heuer

Dr. Daniel Buschek