Visualising Blogs

Jon Schull writes on visualising the relationships between blogs and blog posts (“BlogThreads”).

Jon Udell: My reflex comment is that if the authoring UI were to capture just a sprinkling of metadata — for example, cues that a post intends to “opine” or “clarify” or “disagree” or “summarize” — then these kinds of visualizations would become much more feasible. But the use of such cues, like the use of titles, would take a little time to do, and a little thought to do well.

Dave Winer: As long as I’ve been doing outliners, people have been trying to do boxes-and-arrows visualizations of the same structures, with tantalizing and colorful demos, that aren’t too useful. I did a project myself in the mid-80s. The user interface was unwieldy.

BlogStreet: Quarter Update

When we began, we wanted to launch a blog directory and search engine in a month (by May). We started classifying blogs by hand and did about 200, before we realised that this was not the right approach. Blogs are not as easy to classify as web pages. Besides the manual approach would not take us too far. So, then we stepped back and thought through what we wanted to do.

We still want to make a Blog Directory, but this time around the focus is on (a) the hubs (b) identifying blog clusters (c) doing the categorisation automatically. The two building blocks for this are the blogroll (with the blog links) and the blog posts. So, we are focusing first on doing the blog neighbourhood analysis using the blogrolls and links from the top page of a blog. We have built a system using proglets which identifies blogrolls reasonably accurately.

In this quarter, the plan is to do the following:
– Given a blog, identify its neighbourhood, using what we have called the “Commoner” method: take the most common blogs from all the friends blogrolls and give out a most common list of blogs, in addition to myblog friends, as related. That is if a blog appears among the highest number of times in all friendblog’s blogrolls then it is treated as related.
– Give a rank to each of the blogs, and thus be able to identify the top 100 blogs
– Provide a keyword-based Search engine on the top pages of all the blogs we bot
– Link to a blog’s RSS feed (so it can be sent as an input to an RSS Aggregator)
– Automatically classify blogs by identifying the “hub blogs”
– Look within blogs to identify a Blog Post
– Bot the archives so we can build a search engine with the granularity of a Blog Post (ie, if one searches for a keyword, rather than returning just the page on which it appears as search engines would currently do, it can identify the actual blog post)
– Later: build a live map of the Blog Network (one the lines of what Barabazi discusses in his book “Linked”)

The “hidden agenda” is to be able to leverage the work being done on the external world of blogs to organisations to map out relationships within enterprises as their employees start blogging. (To get them to blog is the goal of the Digital Dashboard project.)

Blog Maps

An interesting idea: bloggers organised by subway station (New York Times): “New York City Bloggers aims to be a comprehensive list of New York blogs, grouped by subway stop…In its first week online, the site attracted 25,000 visitors, and more than 600 bloggers identified themselves by subway stop. (By yesterday, there were more than 1,000.)”

What is missing in the world of blogs are directories. There are a few, but none which take into account that there are people doing the blogging, and what is more important than the blog itself is the person. We need categorisations based on the bloggers, and not just the blog.

A blog is actually very hard to classify, because typically they cover a wide variety of topics. But by looking at the blogroll and the links embedded in the blog, it should be possible to get an indication of the “hub blogs” which the blog relates to. At the same time, bloggers should also “identify” themselves — their likes/dislikes. The two taken together could work well in helping a person find bloggers of interest.


Interview with Slashdot’s Rob Malda: “To a lot of people, Slashdot is nothing but 12 links to new things every day. To half of our readers, in fact, that’s all Slashdot is. But to some of our readers, it’s a community that’s here to discuss issues that are relevant to this community. There is a lot of value. The bulk of our content comes from other people. There are 6,000 or 7,000 comments on a busy day that other people write and just a dozen stories of just a paragraph or two that we actually generate, that are ours.”

I strongly recommend visiting Slashdot daily. Make sure that you create a login and set a threshold for the comments (3 is a good figure). Slashdot was also talked about in Steven Johnson’s book as an example of Emergence — where the whole is greater than the sum of its parts. Watch how people’s comments can help spark new ideas.

Newspaper and Weblogs – Winer

Dave Winer on newspapers and weblogs:

First, I would offer a copy of Radio UserLand to every person on the editorial staff (okay, I’m biased) and say “Start a weblog now if you want.” Then I’d make the same offer to the readers. Then I’d watch to see what happens.

I’d say to the staff “Read the new weblogs, and for those of you who have your own, point to the articles you find interesting or useful.” Let this run for a few months. My bet is that the community starts generating good news reports, on things like school boards, and city council meetings, the stuff that the organizations no longer cover….Just what people see and what they think.

My bet is that the community gets energized by the new participatory journalism and the former reporters, who now are editors, talent scouts and teachers, are also energized, doing what they wanted to do when they got into journalism. Now ask the community what they’re willing to pay to keep the system working and growing. I know I’m naive and unrealistic, but this is how I think it will work.

Another source of revenue. Charge local businesses to place their weblogs on your network. This is advertising turned around.

As always, a very interesting set of ideas. Need to think about them in the context of building “emergent” SME Clusters.

More on Social Networks

Jon Udell on Seeing and Tuning Social Networks: “The growth of weblogs has created a wave of interest in social network analysis….New forms of social software are one of the most hopeful green shoots erupting from a still-bleak technology landscape….As social and computer networks converge, that will inevitably change. We’ll see relationships more clearly and — for better and worse — more analytically.”

There’s been a lot of talk in the blog world of late on social networks. The RSS auto-discovery, blog neighbourbood analysis are two examples of what’s fueling the desire to (a) find other like-minded people (b) find people who can open up other worlds.

A lot of this applies to within the enterprise also. Most of us work in our silos, rarely going beyond the sphere of interest and work. It’s hard to do this today. But if everyone maintained blogs, then this can become so much easier. Knowledge sharing becomes non-intrusive (I don’t have to go to a person to know more about he is working on or what he is good at). By analysing blogrolls, subscriptions and blog posts, it will become possible to see the social networks within enterprises.

This is what has made me re-think on BlogStreet in the past few days. We need to think of it more as a BlogBrain. More on this next week.

Blogging Communities

Matt Mower writes about Creating Communities From Thin Air:

What I have in mind is a program, I haven’t thought of a name so lets just call it blog connector (BC for short), that you can register your blog with. BC then indexes your blog and creates a list of key words that it thinks are relevant to the content on your site. You are then asked to prune, extend and rank the keywords.

Then BC takes a look through all the other blogs it has indexed looking for blogs whose keywords have similar rankings. It checks to see whether you are already linking to those blogs and if not it sends you a suggestion that you might want to check out that blog and/or link to it. It also applies the same criteria to the other blogs, optionally sending them similar information about your blog.

Matt, we are working on just such a thing with BlogStreet. The first part would be a directory and search engine. The second part of it would work on automatically identifying clusters of blogs (and bloggers). Probably hard to do with keywords only — it will need a mix of looking at the blogroll, links and the text in the blogs.