These are my notes for Paul Warren’s talk at the UK Ontology Network Meeting on 14 April, 2016.
What can we learn from the theories of reasoning? To understand the difficulties experienced with DLs, and try to mitigate those difficulties. There have been a lot of psychological studies on how people reason. Historically, there have been two camps, the rule-based and the model-based reasoning. Rule-based (syntactic) is where we reason by constructing logical steps akin to those created formally by a logician in a proof. Model-based is about reasoning via constructing mental models which represent the situation (semantic). These two are complemented by a third method (missed that name).
English: John only has sons (implication: he does have children but no daughters)
Manchester OWL: John has_children only Male (implication: if he has children, they are sons, but he might not have any at all)
English: John sons
Manchester OWL: John son(s), John has things which are sons
It seems that people are reasoning syntactically, based on the work he has done. Recommendations are then to use syntax where possible to emphasize semantics. However, beware of false equivalences and teach the duality laws as expressed in Manchester OWL syntax and let tools show alternative equivalent statements.
In a study, only half got it right that the following were not disjoint: “has_child only MALE” and “has_child only (not MALE)”. Replacing only with only or none helped a lot.
People didn’t do very well thinking about functional object properties. But why was this? He designed two pieces of reasoning which were equivalent which required 3 reasoning steps to get to the appropriate conclusion. One piece used functional, and the other used the transitive property. People finished the reasoning faster and more got it correct when using transitivity.
Functionality is inherently difficult. They suggest a new keyword solely, to emphasise that it is the object which is unique and not the subject. This showed a significant improvement in performance.
Theories of reasoning and language provide insight and lead to recommendations for modifications to syntax, tool support and training.
Please note that this post is merely my notes on the presentation. I may have made mistakes: these notes are not guaranteed to be correct. Unless explicitly stated, they represent neither my opinions nor the opinions of my employers. Any errors you can assume to be mine and not the speaker’s. I’m happy to correct any errors you may spot – just let me know!