Generating MCQs is a time-consuming and error-prone process. Generating them automatically from existing knowledge bases would be desirable, but is an ongoing and largely onsolved problem. In a collaboration with Elsevier, we attempt to develop a novel approach for generating MCQs based on the OWL version of a large clinical knowledge base, EMMeT. Contact: Bijan Parsia
Data related to clinical trials is big, complex, and comes with massive amounts of constantly changing background knowledge. Accessing this data in a flexible, transparent, and reliable way is crucial for companies in life sciences and healthcare. Ontologies promise to capture rich background knowledge, and related Semantic Web (SW) technologies promise to provide the above-mentioned access to this data.
In this project, we explore the potential opportunities, cost and benefits of ontologies and SW technologies for capturing and accessing clinical trials data. We develop an OWL-based model to represent certain aspects of clinical trial data, build a demonstrator that allows a domain expert without OWL knowledge to specify a trial and search for trials, and a further demonstrator that translates clinical trial data between this model and SDTM. Contact: Uli Sattler
NICE provides national guidance and advice to improve health and social care, e.g., through clinical guidelines. Developing and maintaining these guidelines is a time-consuming and difficult task that is currently done manually by groups of health experts using a document-based format. NICE are adopting Semantic Web Technologies for a more maintainable representation of clinical guidelines.
Semantic Web Technologies provide the means to capture knowledge about a domain in machine-processable way and the usage of this knowledge in applications. Recent research on ontology languages and tools has produced results that could be used to support the task of representing and maintaining clinical guidelines. This project aims to develop a demonstrator consisting of a suitable part of the guideline ontology and a demonstrator applications highlighting the benefits of semantic web technology for the development and exploitation of guidelines. Contact: Uli Sattler
Phylogenic trees and Cladograms in OWL: A case study on Dinosaurs
We investigate how ontologies and reasoning can support the design and evaluation of phylogenic trees. In particular, we are interested into logically and conceptually sound models of clades, groups, and specimen, as well as their relationships (e.g., developsFrom) and properties (e.g., genotypical or phenotypical features). Contact: Uli Sattler
Automatic question generation (AQG) is a process that involves using computer technology to generate questions. One of the limitations of the AQG techniques is the simplicity (or even the lack) of difficulty models. Sources underlying the difficulty of questions need to be identified and integrated into the generation process. Thus, the formulation of a theory behind an intelligent automatic question generator that is capable of both generating question of varied difficulty and predicting their difficulty accurately is at the heart of my research project.
Previous PhD theses
General Terminology Induction in Description Logics
Since manual engineering of TBoxes in Description Logics and OWL is a difficult, time-consuming task, automated acquisition of them from data (ABoxes) has attracted research attention and is usually called Ontology Learning. This project investigates the problem from general principles and formulates it as General Terminology Induction aiming at acquiring general, expressive TBox axioms from data (given ABox) while taking available background knowledge (given TBox) into account. We design a semantically sound approach and implement it in a system called DL-Miner.
Capturing Temporal Aspects of Bio-Health Ontologies