Likeness and Relating Similar Items
A newish feature in Frassle indicates that someone else has linked to the same resource with the text of “posts on this subject” based on the link behind the post title. Developer Shimon Rura programmed this feature as a way to link posts that might be similar. One of his goals with Frassle is to connect people together or posts based on similar interests, similar items. It’s a way to facilitate resource and people discovery.
It’s a good start, but it’s not sufficient. I think Shimon acknowledges that. Part of the limitation of it is just the label “# more on subject.” Just because someone points to the same link doesn’t mean what s/he writes is on the same topic. If the language is slightly different, it could better reflect what the similarity means. Maybe “# other person/people point to this link” better reflects how the posts are alike.
Another part of the limitation is just trying to figure out what makes one post like another. It’s something many people grapple with, not just Shimon. I think his attempt to address the problem is quite admirable; and, his efforts to figure out a solution to the problem of finding and connecting similar items is a big step forward along this path. I’ve heard he’s working on a way to evaluate links within the text of posts, too. That could find many more similar items throughout Frassleland.
The other night, an economist talked of a similar problem with journal articles. He wants something much more comprehensive than a citation index, which links journal articles together based on who’s citing whom. He wants to be able to track citations for many articles at a time, like having a citation map, and he wants to be able to look at citations over time, perhaps to measure paradigm shifts. People frequently use those databases to learn about related research by tracing citations. There’s a thought that if someone’s citing an article, what s/he’s writing must be similar. Like connecting blog posts based on what someone links to, it’s faulty because someone can cite an article for many reasons, not just because s/he’s working on something similar.
Of course, if someone, perhaps someone with indexing skills, looked at all of the content in the world, s/he could judge and mark likeness. Making something that does the work automatically is much more powerful, as long as it makes accurate connections. The problem, of course, is that everything’s related based on one person’s or system’s standards for what is related.
Addendum 8/5: Someone shared this quote with me: “Following a reference from a weblog or from a scholarly article are each similar steps in exploring threads of related ideas.” Ariadne ‘Towards Library Groupware with Personalised Link Routing’, Chudnov et al. Issue 40 (July 2004). The problem, of course, is figuring out how to relate those threads in the first place.
