Academic Knowledge Ontologies and a Systems Solution

28th Annual Conference of the Cognitive Science Society & 5th International Conference of the Cognitive Science

Akifumi Tokosumi ; Naoko Matsumoto ; Miho Tomioka ; Klaus Voss

PID  :  123236

来源: CEUR

【 摘 要 】

Two academic areas (a) cognitive science—an established butactively changing field—and (b) largescale knowledgeresources—an emerging discipline—are chosen as targets foracademic domain ontology design research.Among thefindings obtained from designing domain dictionariesreadable for both humans and machines, this paper discusses(a) theinterfacing of local knowledge to global knowledge,(b) the creative nature of academic concepts, and (c) an agent based approach to support the creation of academic concepts.Domains of the Ontologies and Research GoalsIn this paper, we describe findings and insights concerningthe nature of knowledge within academic domains that wehave gained through our work on constructing two academicontologies:(a) cognitive science—an established but actively changingfield, and (b) largescale knowledge resources—anemerging discipline.The goals of our ontology construction work, conceivedof as a human and machine readable collection of lexicalitems and concepts, are:i) to support communication between researchers in relatedfields,ii) to support new students, andiii) to support translation.We summarize our findings into the characteristics ofacademic knowledge.Locality and Globality of Academic KnowledgeTheory dependency is an especially prominent feature ofconcepts within the two areas under consideration.While itis almost impossible to provide a single universal definitionfor any concept, we found that it is possible to find workingdefinition for practical purposes, once a theoreticalperspective is identified.Because, in general, scientificpapers include a large number of implicit assumptions thatdepend on a particular theoretical viewpoint and socialbackground which are rarely ever stated explicitly, itrequires domain expertise to find a suitable definition for thefield (Tokosumi, et al., 2006).This is especially problematic in two typical cases: (i)communication between researchers in neighboring fieldswhere shared knowledge may not be sufficient, and (ii) innew students.Analyzing both (i') the interdisciplinary gap,and (ii') the expertise gap, we have found that both can bereduced in terms of the locality of academic concepts.The Private and Public Nature of OntologiesIt is essential for an academic ontology to appropriatelyfactor in the bidirectional nature of the concept revisionprocess.Scientists are, by definition, innovators of concepts.The process of proposing new or updated concepts varies atseveral levels, such as the individual researcher, thelaboratory, scientific communities, and scientific journals. In this process, concept definitions need to be circulatedsimultaneously within subsets of the scientific communityand within groups of individual researchers.In this paper we address this process by proposing aframework to support the creation of both private user andcommon groupdictionaries or docu