Wiki Link Structure Analysis

Consider trying to find WalledGardens, CulDeSacs and other interesting bits. The idea is to identify those pages that may contain valuable information, but which are not appropriately linked to (or from) related pages. Such pages may prove hard to find simply because they are not sufficiently linked to, or it may be that having found a valuable page you can't find related pages because they have no outgoing links.

Thus we are looking for The script at http://c2.com/cgi/orphanWikiPages which does something similar. That script, however, does not identify pages pointed to only by orphans, nor does it identify orphan clusters.

Please note that the following are not value judgements. These are simply statistical truths about the link structure. The pages may be valuable, they may be dross. This analysis does not show that.

The problem is that sometimes there are clusters of pages that exist in isolation from the main body of the wiki. Of itself this probably doesn't matter, although they may be WalledGardens or examples of WikiSquatting. The potential problem is that someone may come along and contribute more pages, duplicating those that do exist but are hard to find.

This analysis is therefore intended as a tool to help WikiGnomes and WikiZens to find and subsequently link together related pages. Perhaps it will prove useful, and that some otherwise overlooked gems may be found and made more widely known.


Here is the algorithm: Performing the above analysis (April 2008) shows that there are clusters that are both orphans and dead ends. For example:

There are currently (April 2008) 132 orphan clusters and 81 dead end clusters, the 16 above are on both lists. These figures are up from 123, 77 and 14 two years ago.

You can check whether a given page is on one of those lists by simply hitting "Find text" in your browser (or type the page name, if you're using MozillaFirefox with find-as-you-type).

See also OrphanClusterList and DeadEndClusterList
CategoryAutoIgnore

EditText of this page (last edited January 2, 2011) or FindPage with title or text search