• Macquarie Home »
• Faculty of Science »
• Department of Computing

Abduction for text interpretation

Abduction, or inference to the best explanation, is a method of reasoning that starts from a given special case and a known generalisation and then makes a guess that explains the special case, for example:

Given: Tweety flies.

Known: Every bird flies.

Guess: Tweety is a bird.

In this honours project you will investigate how a description logic reasoner can be used to generate explanations for a specific application domain (of your choice). This requires the construction of a description logic knowledge base that combines facts, terminological knowledge and rules (= known generalisations). The idea is to extract facts automatically from unrestricted but domain specific texts using an existing shallow text processing tool and to create the terminological knowledge and the required rules for this domain (in a frist step) by hand. The extracted facts will be divided into two parts: a part (a) that the user would like to have explained and a part (b) that the user takes for granted. The facts of part (a) are then transformed into corresponding queries and the abductive reasoning service is asked for explanations. The goal is to implement a protoype of such an abductive reasoning service that uses a state-of-the-art reasoner as backend.