Faculty of Computer Science

Research Group Theoretical Computer Science

Oberseminar: Heterogene formale Methoden

Abstract:

The Human Behaviour-Change Project is a large-scale cross-disciplinary effort aiming to create a platform for automating the synthesis and integration of evidence relating to human behaviour change interventions directly from the published literature. The project has multiple components: an ontology for the controlled description of behaviour change interventions and the studies that evaluate them, a corpus of manual annotations of the published literature on intervention trials, an information extraction system that aims to automate the annotation of newly published literature, and a prediction system that learns from the annotated literature to predict outcomes based on specified parameters. Although the project uses an ontology as input for the annotations, the semantics of the ontology is not used in the current prediction system. In this presentation we will discuss the architecture of the prediction system, some challenges that it faces, and how the addition of ontology-based semantics might address those challenges.