- From: Mark William Mitchell <mwm8d@cms.mail.virginia.edu>
- Date: Mon, 21 Oct 2002 11:37:36 -0400
- To: www-rdf-interest@w3.org
Hey everyone, I have been toying around with an idea lately and would like to get some feedback. I am writing my senior thesis basically about how to advance the adoption of the semantic web. It seems to me that all the tools exist already that are necessary for the semantic web to become a reality, but that we are still a long way off from making even, say, 10% of the data on the Internet machine-understandable. The biggest stumbling block to the adoption might actually be a human interface problem. Web authors do not even take the time to fill in a few simple metadata fields to make their sites more search engine friendly. A lot of the time, metadata is gathered after a document has been written, either by semantically limited automated tools, or by functional experts combing through tons of documents. Knowledge bases usually only cover very specific domains and are not linked to any data external to their own. If we are ever to have large quantities of machine-understandable data, capturing metadata at a document's creation time is critical. What about an application that collects metadata from a document at the time of its creation? Here's a possible scenario of use: A consulting company wishes to build a "Lessons-Learned" knowledge base from many of their software development projects. Developers would simply write down a "lesson-learned" in say, Word, or in a text file, or documented the problem in a piece of code. They then would not be constrained to entering information into a few, already defined, database fields. They could have a graphical representation of the lesson-learned, or a code sample, or a URL to a helpful hint, etc. To create a "lesson-learned", the developer would create whatever file he wished to represent the information in, and save the file in his "lesson-learned" directory. The application I am talking about would now pop up a dialog with common metadata fields pertaining to "lessons-learned" and parse through the file (if it knew the file type (such as straight text)). The developer would enter in a few fields, and boom, good metadata about a pre-defined semantical file type. This information could then be centralized on a server or searched in peer-to-peer fashion by other members of his compnay/team. An application that knows what type of "semantical file" you are creating would be very useful. By "semantical file" I mean like say, an "Architecture Drawing". It is semantical becuase it is a concept. It could be a JPEG, GIF, some sort of cad file, PDF, etc. That would be what I call its "syntactical" file type. The group you are a part of would have a functional expert define a few metadata fields relating to the semantical meaning of your document, and then the document creator would easily fill in the appropriate information at creation time. The metadata from this would be shared across the network. There are obviously a lot of grey areas as to how-to implement this, but I don't think it would be too difficult. The main problem is not to invent some new protocol, but how to share the metadata gathered in a standards based way that other applications would have no problem interfacing with. Any thoughts? Thank you for your time, Mark
Received on Monday, 21 October 2002 11:37:42 UTC