Talk: Silk Purses from Sow’s Ears: Generating Quality Multiple Choice Questions from Ontologies

Printer-friendly version

When: September 15 2014
Where: Winslow Building Room 1140, RPI Campus, Troy, NY
Dr. Bijan Parsia will be visiting the Tetherless World Constellation lab next week. He will be giving a talk on Monday, September 15 at 1:00pm in Winslow, Room 1140.

Short Biography
Bijan Parsia is a senior lecturer at the University of Manchester (UK) in the School of Computer Science. Bijan Parsia is a member of the Information Management Group (IMG) since May of 2006. Before that, he was a Faculty Research Associate in the Mindswap group at the University of Maryland College Park (UMCP).

Title: Silk Purses from Sow’s Ears: Generating Quality Multiple Choice Questions from Ontologies

Abstract: One promise of Knowledge Representation (KR) is that a properly designed, representationally adequate KR in a sufficiently powerful language will be reusable in many contexts for many applications beyond the original intended purpose. The classic example of this was the attempt to repurpose the early diagnostic expert system, MYCIN, into a tutoring tool (GUIDON). The key hope was that the knowledge encoded in MYCIN which was sufficient for *performing* diagnosis would be useful for *teaching* how to diagnose.

Today, the Resource Description Framework Schema (RDFS) language and the Web Ontology Language (OWL 2) are probably the most popular logic based KR languages extent with thousands of ontologies and datasets availed for a wide variety of topics with varying degrees of complexity and comprehensiveness of representation. The BioPortal community repository has a growing, actively evolving set of hundreds of ontologies focused on the life sciences. This constitutes a vast reservoir of knowledge to reuse.

In this talk, I will discuss a small piece of this puzzle, the generation of high quality Multiple Choice Questions (MCQ) from OWL ontologies. To this end, I first discuss the basic theory of MCQs, their utility, and their quality dimensions. Then I will propose a psychological grounded theory of MCQ difficulty (a key metric) base on similarity theory. Then I will discuss how we can interpret similarity in the context of logic based ontologies. Finally, I will discuss some experiments and experiences in applying all this to generating classroom ready MCQs.

Host: Jim Hendler