Udskriv

Analysis of Subjective Conceptualizations Towards Collective Conceptual Modelling

Publikation: Forskning - peer reviewKonferencebidrag i proceedings

This work is conducted as a preliminary study for a project where individuals' conceptualizations of domain knowledge will thoroughly be analyzed across 150 subjects from 6 countries. The project aims at investigating how humans' conceptualizations dier according to dierent types of mother languages, cultural backgrounds, gender, generations, etc., when domain-specic terms are expressed in a common language, i.e. English. In this work, we analyze a publicly available dataset [De Deyne, 2008] representing semantic structures of domain knowledge possessed by four subjects. The application of a non-parametric relational model, Innite Relational Model [Kemp, 2006] co-clusters concept-feature relations, which identies a common semantic structural grid across the four subjects considered. Through this common grid, the individual semantic structures possessed by the respective subjects are further contrasted to each other. The results show that the identied common grid visualizes subject specic patterns among the extracted concept-feature clusters. The work further discusses our future perspectives for modeling a collective ontology across subjects according to dierent subjects' proles.

Publikationsoplysninger

OriginalsprogEngelsk
TitelProceedings of the 27th Annual Conference of the Japanese Society for Artificial Intelligence (JSAI 2013) : International Organized Session: Modern Approaches for Intelligence Design. From Mining to Inference
Antal sider9
Udgivelses stedwww
UdgiverJapanese Society for Artificial Intelligence
Publikationsdato2013
KapitelPaper 795
StatusUdgivet - 2013
Scopus citationer
BegivenhedThe 27th Annual Conference of the Japanese Society for Artificial Intelligence (JSAI 2013) - Toyama, Japan

Konference

KonferenceThe 27th Annual Conference of the Japanese Society for Artificial Intelligence (JSAI 2013)
Nummer27
LokationToyama International Conference Center
LandJapan
ByToyama
Periode04/06/201307/06/2013
Internetadresse

ID: 38655398