Re: [TaxoCoP] Hierarchical relationships [Was visualization technologies]
Many thanks for your comments on qualified relationships. I would have responded sooner had I not been having connectivity problems and limitations here in Europe.
I accept entirely your rationale for "reducing thesaurus relationships to the basic three" in the context of information retrieval. However, ever since I became acquainted with knowledge management, I have had a bit of a problem with the term 'information retrieval'. Unless I've got the etymology completely wrong, it means 'to find again', implying that it is a process of getting back something you had formerly. It seems to leave out of account the process of discovering something you hadn't had before and didn't know existed. I therefore like to distinguish between 'retrieval' and 'discovery', although both processes are important for learning and innovation.
When referring to the elaboration of relationships and entity types in the CIDOC CRM, you ask "what are you going to do with it?". I think it is in the process of discovery that qualified relationships can be useful. Although RTs in thesauri allow some escape from one hierarchical structure to another, it seems to me that one is still confined to a two-dimensional surface when effective exploration and discovery requires a three-dimensional space, which I believe the qualification of RTs provides.
I see the qualification of relationships in the vertical (intensive) dimension as supporting quite different outcomes from the qualification of relationships in the horizontal (extensive) dimension. Greater intensivity is useful for subject specialists in delving ever deeper into aspects of their specific domains, and it is largely this process that has produced the body of scientific knowledge from which modern society (sometimes arguably!) benefits today. But there are those who claim that in post-industrial society, innovation arises increasingly not from more intensive knowledge, but from more extensive knowledge, from the combination of insights from entirely different domains. One of the major themes of knowledge management is therefore to enable and support this serendipitous form of innovative discovery.
If this is the case, then the better we can express the totality of the known context of an entity or concept, then the greater the support we lend to the discovery process (and, incidentally, the discovery of hitherto unsuspected relationships). The qualification of horizontal relationships allows one to build a network of semantic relationships which provides a more realistic model of the world and helps explorers to move beyond the boundaries of a single discipline or focus and to make connections, build understanding and enhance learning.
I think there is a popular perception that innovation occurs solely within a single discipline, which has arisen from the practically-based but artificial distinctions among disciplines. Anyone who has studied physics and chemistry for example, is aware of how the two are inextricably interlinked. Yet we persist in treating and in teaching them as quite separate disciplines, even though most technological innovations in the last fifty years derive from a combination of the two. Silicon chips, for example, are the result of innovation in both physics and chemistry.
I agree that "We have to specify the relationships more explicitly if reasoning is to be done by machines rather than by humans" and I believe the imperative to utilize computers in doing some of our reasoning for us is what is driving the information science and AI/Knowledge Engineering communities to converge. I think one of the most flexible and sophisticated expressions of this convergence is the XTM Topic Maps standard, ISO 13250. Topic Maps provide not only for characterization of the relationships (it calls them 'Associations') among entities, but also for the definition of the role of each entity in the Association. I think the Topic Maps codification standard is what will allow us, at last, to enlist the true power of computer processing in extracting value from the glut of information which currently confronts us.
Hmmm. I think I'd better get down from my soapbox now, and give others a chance to comment :-)