Whenever you write a blog post that contains links to other posts, and is maybe in turn linked to from other blog posts, how can you keep track of where you blog post sits “in the wider scheme of things”?
In Trackforward – Following the Consequences with N’th Order Trackbacks, I showed a technique for tracking the posts that link to a particular URI, the posts that link to those posts and so on, suggesting a way of keeping track of any conversational threads that are started by a particular post. (This is also related to OUseful Info: Trackback Graphs and Blog Categories.)
In this post, I’ll try to generalise that thinking a little more to see if there’s anything we might learn by exploring that part of the “linkgraph” in the immediate vicinity of a particular URI. I’m not sure where this will go, so I’ve built in the possibility of spreading this thought over several posts.
So to begin with, imagine I write a post (POST) that contains links to three other posts (POST1, POST2, POST3). (Graphs are plotted using Ajax/Graphviz.)
In turn, two posts (POSTA, POSTB) might link back to my post:
So by looking at the links from my post to other posts, and looking at trackbacks to my post (or using the link: search limit applied to the URI of my post on a search engine) I can locate my post in its immediate “link neighbourhood”:
Now it might be that I want to track the posts that refer to posts that referred to my post (which is what the trackforward demo explored).
You might also be interested in seeing what else the posts that have referred to my original post have linked to:
Another possibility is tracking posts that refer to posts that I referred to:
It might be that one of those posts also refers to my post:
So what…. so I need to take a break now – more in a later post…
See also: Tweetbacks, a beta service that provides a trackback like service from tweets that reference a particular URL.
PS and also BackType and BackTweets
OUseful.Info, the blog... 
Is this basically what PostRank does?
I like the direction this is traveling in, but the output could be mighty confusing. Ultimately, I’m going to need it presented as some sort of WordPress/PostRank-style dashboard.
Yes – I think there are similarities to PostRank; what i want to think towards, though, is a conversation tracker and navigator…
I have no idea where this line of thinking might go either… which is what makes it interesting, right?;-)
I like it…things like this should be relatively straightforward to build, at least for the early stages, from an RDF graph (after all, you’re pretty much drawing RDF graphs in your diagrams!).
The trickier part, as ajcann gets at, is going to be how to represent the output.
I’ll dig through some of the linkage data I’ve got from umwblogs.org and see if there’s a useful graph I can grab out of that for tinkering and poking around the interesting edges of where this might go.
Sounds like citation analysis for journal articles, but for blog posts instead. Web of Science allows you to analyse citations within a set of results using their Citation Analyser and that includes some graphs – may be useful to see what their approach is :)
@clari That’s a good suggestion. There’s been a lot of research over the years on citation schemes. It might be beneficial to transfer some models to RDF formats.
One thing that intrigues me is just the geeky practicalities of how to get the links tracked in the first place (at least well enough for a crude proof concept).
At the simplest level, I guess I just want to find edges between nodes, even if those nodes/edges have a different ontological status (e.g. post hoc references back to a post might be made from another blog post linking/tracking back, a comment to the orgiinal post (which in turn might contain a link elsewhere), a tweet that links to tthe post via a tinyurl etc etc; and i suppose I want to retain the direction of those edges?
Re the parallels with citation analysis – yes; PageRank is famously inspired by that sort of approach, of course;
@patrick your linkage data would be interesting to see; Stephen Downes also expressed some interest in mapping the link structure of edublogs last year – he may well have a huge corpus of blog posts we can play with, maybe using oldaily as a catalogue of seeds that can be used to start to crystallise the link structure out?
Okay, here’s another thought – suppose we actually resolve post/tweets/comments back to their authors? What sort of graph do we get then? One that defines a community, I guess, But maybe also one that helps us pick out “discussion types” or “discussion signatures”, e.g. in the sense of flashmeeting analyses that identify different video meeting types (discussion, lectures etc etc) based on who participates when and for how long – http://flashmeeting.open.ac.uk/research/shapes.html
So if in s conversation graph, I’m the only contributor, linking to myself, I’m maybe developing an argument or line of thinking; whereas if there is turn taking going on bewteen particpants, there’s more of a discussion going on maybe?
Very cool idea about deriving types from the structure of the graph!
Here‘s a bird’s-eye-view of the linkage graphs from UMWBlogs
See also:
BackType http://www.backtype.com/
BackTweets http://www.backtweets.com/
@Tony: Your second diagram basically illustrates how BackType Connect works (although Digg, Reddit, aren’t looked at this way):
http://www.backtype.com/connect
We didn’t include outbound links because (1) those are already discoverable, and (2) the conversations became noisy. Another important distinction is that we won’t include a node if it doesn’t have any conversation.
Here’s a good example of Connect in action:
Thanks for your interest in BT!