Weblogs: Semantic Web

Bill Kearney on Syndicating Topics

Monday, July 14, 2003

Bill Kearney is thinking about a synthesis of people and the topics or subject matter they contribute to.

The approach he wants to take is to identify a topic, and from that determine who is in the know about that particular topic. Naturally, the selection process revolves around trust- or reputation-based systems, and there is a problem of abusive pollution of these spaces.

Google offers a possible solution in that every link to a URL is a vote for its trustworthiness, so that oft-linked pages are deemed to be authorative or good enough for a top-listing on Google. It also highlights the pollution problem, when searching for the term "weapons of mass destruction" in Google currently returns the current top result of a humourous mixture of the WMD contraversy and a MSIE error page.

Directory-based (human edited) systems are safer in this respect - pulling the wool over the eyes of a human being is a lot trickier than a search engine indexer. The drawbacks of a human-edited directory is that it enforces the "super-editor" status on a handful of people - opening up another avenue of abuse, and leading to the sort of political wasteland Bill Kearney is keen to avoid.

But its probably in a directory based system that the solution to this problem lies. Except the directory editor must not have a self-interest in the topic at hand (especially a commercial interest). Most likely a counter-balance needs to be introduced, allowing participants (already known topic-experts) in a topic to "cast a vote" at the merits of a new resource, and even an introductory boost in value of new contributions from new people - like a shooting star - which lets the topic-experts evaluate the new contribution for value-weighting after the introductory period has finished.

Google have used these techniques over the couse of its indexing - Page Ranking is a well known Google page valuing technique and link-karma sharing, and currently they seem to boost the page rankings of new pages (probably a factor in the blog dominance of search engine results).

The problem with this sort of system is that it is difficult for a new page to be recognised. Maybe some sort of topic-based or author-based trackback is the initial push needed to identify a new contribution - although somehow somewhere someone needs to "authenticate" it by either linking to it or recommending it.

The combination of character and topic based space is an interesting idea - even to the extent of a person and topic combination allows the discovery of other areas of interest and finding more things in common with a particular person. Its an interesting idea, and something worth pondering on over the next few days.


[ Weblog | Categories and feeds | 2011 | 2010 | 2009 | 2008 | 2007 | 2006 | 2005 | 2004 | 2003 | 2002 ]