Say you have an 'item' that you really like. How do you go about finding more, similar items?
1) Follow citation links
a) Items that this item cited (references), and so on.
b) Items that cited this item ('cited by' links, trackbacks, Google's 'link:', Technorati, CiteSeer), and so on.
c) Items that this item cited were also cited by ...
d) Items that cited this item also cited ...
2) Analysis of the item itself
a) Text analysis - shared keywords, latent semantic indexing, etc
3) Collaborative filtering
a) People who ... this item also ...
Like/Dislike (Soulseek)
Listened to (Audioscrobbler)
Bought (Amazon)
Bookmarked (del.icio.us)
Listed (Amazon)
Published (record labels)
Stored (browsing shared P2P folders)
Rated (reviewers)
4) Ranked relations.
a) Take the closest related items, and rank each of their related items. Combine the scores to produce an even closer cluster of related items.
b) Display the related items visually, as a graph, and pick out clusters and shared links.