Finally! Bloglines Blog Search

Ask.com, which owns the most popular stand alone web based feed reader, Bloglines, has just rolled out its long awaited new blog search engine.

Ask/Bloglines has been the subject of a considerable number of jokes over the last year, after promising a blog search engine last summer. The new engine should put those jokes to rest. And the company is taking the product and the launch very seriously – Ask.com CEO Jim Lanzone walked me through the product personally earlier this week.

The search engine has two separate user interfaces. It can be found on the Ask.com home page (link to blog search on the right sidebar) as well as Bloglines. The underlying engine is the same, although the interface and functionality is slightly different on the two sites.

Ask.com Blog Search

Searches can be conducted by “posts”, “feeds” or “news”. The news option conducts a search from 7,000 pre-approved blog and news sites to reduce noise.

Results can be narrowed to a specific period of time (anytime, last hour, last week, etc) and can be sorted by relevance, date or popularity. “Popularity” is determined based on the their “ExpertRank” algorithm and several sources of Bloglines data, such as subscriber count, links, citations, etc. A blog with more links and more subscribers on Bloglines will have more relevance than other blogs. “Relevance” factors in both popularity and freshness to give meaningful recent results.

There are a number of other features worth noting. Feeds related to the query are listed on the right sidebar, along with RSS information for subscriptions. Each search result contains additional options as well: a binoculars graphic (scroll over for popup with last five posts from result), “Save” (save result to a clipboard), Subscribe (to a feed reader) and Post To (Digg, Delicous, Newsvine etc.).

Advanced search features are accessed via an javascript drop down menu at the top of the screen.

Bloglines Blog Search

Bloglines is using the same back end search engine as Ask.com, although the interface and feature set has notable differences. A key feature is a “+” button next to each result. Click on the button and the full post is presented with original formatting (not quite the original formatting actually, but pretty close).

Another difference – each result has a “more info” link that shows the number of bloglines subscribers for that blog and any citations for that post.

My Thoughts on Relevance:

There is a big need for the equivalent of Google Page Rank for blog search relevance. Link analysis on a post just doesn’t work – the content is too fresh to develop meaningful link analysis results. There are now three experiments going on with relevance: Technorati bases relevance solely on “authority” of the blog, which is calculated solely on unique inbound links to the blog itself. This works much of the time, but can break quite easily. TechCrunch, for example, can be the highest rated blog on just about whatever I write about, regardless of whether I know anything about it. If I write a blog post on a political issue, for example, it will appear at the top of results even though I have no qualifications for doing so. Bloglines is taking a different approach, by factoring in a number of statistics such as Bloglines subscribers, link analysis and other information. This may eliminate or reduce the non-topic-specific Technorati authority problem. Sphere is making an effort to assign real authority to a blog on a given topic. They look at links in and out, as well as a semantic analysis of the blog itself. Theoretically, on Sphere a blog that is relevant in one area won’t be relevant in another. It’s a good theory and may work over time in practice as well. But the blog search relevance battle is far from over, and I look forward to new experiments over time.