MEMBERSHIP AND TYPICALITY IN CONCEPT REPRESENTATION: FROM COGNITIVE PSYCHOLOGY TO INFORMATION TECHNOLOGY

The thesis concerns knowledge representation in humans and machines. In particular it focuses on the role of concepts in knowledge representation, a topic at the intersection of Cognitive Psychology (CP) and Information Technology (IT). When humans and machines need to interact, problem dependent on different mechanisms for representing the same knowledge emerge. This issue is broadly debated in the recent literature. An optimal interaction between humans and machines could be eventually achieved by taking into account the human cognitive side of knowledge representation and by making these computational representations cognitively plausible for individuals. The thesis focus on Membership and Typicality in human categorization and takes into account the role that such factors could assume in concept representation in IT, by analyzing their impact on categorization in Web ontologies. The thesis is structured into a first part that describes the specific theoretical contributions of CP and IT, emphasizing the commonalities between the two perspectives, and a second empirical part that reports six original studies, five laboratory experiments and an online survey. Laboratory experiments were based on sentence verification tasks performed by participants, where Membership and Typicality were directly contrasted, with the goal of measuring the effect of such factors on categorization. The online survey explore users' attitudes and opinions towards schema.org, a general ontology for the Semantic Web. Findings are consistent with the idea that – in addition to Membership –Typicality should be considered in concept representation and supports the conclusion that, to be more usable, Information Technology should prefer cognitively plausible ontologies.