Text2Graph - Extracting Knowledge Graph Relations from Short Texts - Einzelansicht

Modern automated text generators (like GPT-3) produce natural language sentences, but often these sentences make no sense or are just plain  wrong, as these generators are still lacking knowledge about the world. The most promising model of world knowledge today to cope with this problem are knowledge graphs, with entities as nodes and named relations (is-a, has-occupation, ...) between them.

These knowledge graphs are often built by machines and have gaping holes in their actual knowlede. In this project we want to extract the missing knowledge from an abundant yet untapped source: internet short texts like microblogs or forum posts. Participants will investigate existing approaches and learn about both natural language processing and information representation.

Time and place will be announced at the project fair.

Abschlusspräsentation und Ausarbeitung

B.Sc. Medieninformatik / Informatik

M.Sc. Computer Science for Digital Media

M.Sc. Human-Computer Interaction

M.Sc. Digital Engineering