ClickAider

Searching for blogs related to a topic

Just about every blog search engine does a decent job of returning recent posts that contain a particular keyword, sorted by the overall popularity of the blog, freshness, etc.

But what if I want to find a blog most relevant to a particular topic? Every blog search engine does poorly on this task, including the new “blogsearch.google.com”:http://blogsearch.google.com.

Suppose, for instance, that I wanted to find out the best recruiting blogs.

“Technorati’s search results”:http://www.technorati.com/search/recruiting give me fresh postings that happen to mention recruiting, but does nothing to help me find the best blogs about recruiting.

What about Google, the king of relevance? “Google’s results”:http://blogsearch.google.com/blogsearch?hl=en&q=recruiting&btnG=Search+Blogs do in fact include a list of related blogs at the top of the page. It’s the right idea, but their implementation leaves much to be desired. The list is far from complete or authoritative– it has only five blogs of no special relevance.

As I’ve grumbled in the past, the problem is that search engines miss the forest for the trees– they treat a blog as a collection of individual postings rather than as a whole from which a unified theme can be inferred. Fixing this shortcoming is important not only for end user but also advertisers… If you can infer a theme for a blog, you can do a much better job of targeting advertisements that are likely to be of interest to the readers. It’s just too easy to get a false keyword hit on an individual posting and end up with advertisements that detract from the blog content.

No doubt Google intends to improve on the relevance of their related blogs feature over time. I really do think that finding blogs related to a particular topic is a key end user scenario, and that the ability of companies like Technorati to survive depends on beating Google at that game.

2 Comments so far
Leave a comment

Hi Phil

great point…

Does the Nooked RSS Directory not achieve some of what you mention

obviously they could do more work to reach the goal you state - but i think a mixture of feed approval, rich meta-data and community participation to drive these 2 items, then they maybe onto something

regards
Bill

Directories and folksonomies like Nooked rely on manual effort and community to build their directory, much as Yahoo’s original hierarchical directory of the web did.

The challenge with manual directories, of course, is making them complete, reliable, and up to date.

Today, when I searched for my sample topic “recruiting” on Nooked, the responses were not complete and perhaps not even quite as good as Google’s related blogs link.

It’s not inconceivable that down the road a community based approach could beat a purely automated one.

There are important differences between web and blog search. With web search, new pages are created at such a pace that a manual directory is doomed to incompleteness. Blogs are created at a slower pace so a directory is potentially more feasible.

The catch is solving the authorativeness vs spam problem for blogs. There will be a great incentive for spammers to spam the directory if it catches on. Getting the communities to invest enough effort to beat out the spammers could be a challenge.

Conceivably you could combine the approaches– use a community to define the hierarchy, use links to rate authorativeness and relevance.


Leave a comment

(required)

(required)