Let me mention that the approach has a good foundation:
Basically information means selection within a domain (value or definition set). For comparable information the domain should be the same for all. Strictly speaking for objectivity and precision of information the uniform definition of the domain is even precondition. Therefore the global (online) definition of the domain is proposed here. It is advantageous to define an ordered domain, because this allows using numbers for addressing the elements and because nature is ordered in many respects. If the original data are ordered in multiple independent ways, we can define a domain with multiple independent numeric dimensions to reflect this. Because we want to search information in the domain, for quantification of similarity we can define a distance function or metric. Therefore we propose “Domain Spaces” (“DSs”: online defined nestable metric spaces, http://numericsearch.com/DS-Page-med.pdf ). With this searchable information can be represented in simple form as “Domain Vectors” (DVs):
URL (of common DS definition) plus sequence of numbers
Compared to words there are important advantages (objectivity, information content and concentration, range, resolution, efficiency etc.). DSs can make quantitative data searchable. They can make generally user defined information searchable. They can be used for globally uniform definition of complex information, e.g. medical findings ( http://numericsearch.com/FutureMedPoster.pdf ) . Of course we need a Web standard for DVs and DSs definitions. Though there are many possibilities for extensions, e.g. possibilities to reuse existing definitions in new definitions, we can try to make the basics as simple as possible. We can discuss this here.
Perhaps more real data for the demo search engine http://numericsearch.com could better show the potential of the concept. If you have quantitative data which should be made searchable and published, please contact me.
Many thanks to all supporters of this group. There has been a long incubation period, but I hope we can now begin. Let me first explain why I have made this proposal: Quantitative data are very important. It would be useful (e.g. for decision support) to make quantitative data searchable on the web. So we need a data structure for identified and searchable (multidimensional) quantitative data. Metric spaces are a natural means for this because all elements of a metric space have a well defined distance. So it is possible to sort the search result according to distance (similarity search). Moreover it is possible to nest definitions (reuse existing definitions as parts of new definitions).
In http://numericsearch.com/wwdomainspaces.pdf nestable metric spaces are called “Domain Spaces” and described in detail. They can be defined by all web users according to their domains of interest. I propose this to start a discussion about the best approach. Feel free to ask questions and to make suggestions. Everyone who is interested is welcome to join the group.
This is a community initiative. This group was originally proposed on 2013-04-26 by Wolfgang Orthuber. The following people supported its creation: Wolfgang Orthuber, amitedu amit, amit asthana, Paul Caesar, Ethan Dagner. W3C’s hosting of this group does not imply endorsement of its activities.