Hi Jonathan, >** The problem ** >On the Web, DC.description and DC.subject are not very effective finding aids >when the full text is indexed.
I'm unclear as to the purpose of your enquiry. My take on what you have outlined is that you're seeking a method of generating metadata records without requiring the author to be involved. If this were the approach taken by the White House, then George W Bush's biography would be assigned the metadata record 'miserable failure'. < http://news.bbc.co.uk/2/hi/americas/3298443.stm > The benefit of classification by an authority (someone who knows their field) is that the classification differentiates content. The more specific the classification, the more useful that classification is to a knowledgeable searcher. On the web, broad, common language classification systems are of most value when the subject is unknown. For example, as a new web designer I might search for 'web design'. As my knowledge increases, my search is likely to be become more sophisticated, for example 'CSS floats' or 'IE box-model hack'. It would be helpful to define both 'effective' and 'finding aid'. As search is such a broad topic it would also be productive to establish context. For example, is this a public search or a site specific search? Would metadata records be displayed to the user or factored into the page ranking? Etc. >** The solution ** >Wild metadata, such as anchor text, blog descriptions and folksonomies may >provide better description and subject (or keyword) metadata. If the author-generated metadata records are displayed as part of a search result records, then they provide a succinct description of the content. As to whether an individual finds metadata record support the locating of content, the method of display, relevance of the metadata records to the search conducted, personal preference, etc also come into play. Link text (i.e. the text used to link to one webpage from another) is already factored in public search engine ranking algorithms, as does the number of incoming links. Trackbacks < http://www.motive.co.nz/glossary/trackback.php > are an existing method of capitalising on blog comments, as the link text and blurb from referring webpage is embedded on the source webpage. With regard to folksonomies, looking through Technorati's tags < http://www.technorati.com/tags/ >, content is often classified according to subjective qualities such as 'rant', 'rambling' and 'random'. It is more likely that folksonomies constitute a snapshot of the evolution of language. As a 'fringe' term become socialised it emerges as part of a formal classification system. For example the term 'hack', as it pertains to CSS, has been socialised to the point where it has become meaningful search term. I would go so far as to suggest that public search engines have already implemented a 'wild metadata' approach to generating search results. Perhaps the issue with the value of metadata records lies less with how they are generated and more with how people phrase search queries and use the web. You might find it useful to browse our glossary as it provides further info on search engines, folksonomies, metadata, etc. < http://www.motive.co.nz/glossary >. Best regards, -- Andy Kirkwood | Creative Director Motive | web.design.integrity http://www.motive.co.nz ph: (04) 3 800 800 fx: (04) 970 9693 mob: 021 369 693 93 Rintoul St, Newtown PO Box 7150, Wellington South, New Zealand ****************************************************** The discussion list for http://webstandardsgroup.org/ See http://webstandardsgroup.org/mail/guidelines.cfm for some hints on posting to the list & getting help ******************************************************