Name of the participant: Tim Steuer
Description of the IT-research project:: We acquire a large part of our knowledge through reading and dealing with texts is ubiquitous in a digital world. The vast majority of textbooks are available in digital form and easily accessible via the internet. Today, it sometimes takes only seconds from initial interest to looking at a textbook. However, quick access to specialised texts does not necessarily lead to efficient acquisition of specialised knowledge. Often, during passive reading, we only absorb fragments of the actual information in the text or grasp the core of the text only superficially. Most textbooks lack a wealth of questions that allow the learner to actively engage with the content.
In this project, we are therefore investigating the extent to which we can automatically ask meaningful questions about the content of any textbook. The challenge of the project lies in information extraction and text generation for the complex and extensive texts. First, we identify important information and convert it into question form. Not every sentence is suitable for a question and not every question that can be asked is didactically useful. We combine methods from artificial intelligence for information extraction and text generation to develop a human-in-the-loop prototype. This automatically suggests possible questions to authors for their text while they are writing the books. This way, authors can incorporate these questions directly into their textbook or still make adjustments.
With the help of this prototype, we evaluate the impact of the generated questions on the authoring process as well as on the learners’ understanding.
Software Campus partners: TU Darmstadt, Holtzbrinck Publishing Group
Implementation period: 01.01.2021 – 01.10.2022
