Apr
Three weeks ago I started my experiment to see if Biznar was finding documents related to the phrase “federated search” that Google wasn’t finding. In that experiment I was shocked to find that, of the 21 results Biznar Alerts sent me on March 14th, only 4 of them had ever been presented by Google Alerts.
Last week, two weeks after the initial experiment, I checked the same 21 Biznar results to see if Google did find them, but just later. You can see the 21 results I used for my experiment plus some clarifications of how I did my experiment in the original article.
What did I find? Google did not find any of the other results that Biznar found. There are two possible explanations for this:
- Google isn’t indexing the sites hosting the 17 results it didn’t find. That’s not very likely. In fact, if I search Google for the titles of a few of the Biznar alerts (as phrases) they do turn up. So, Google isn’t failing to crawl them all.
- Google is deciding not to include them in their alerts. The question is “why?” I imagine that Google has to decide which search results are the most relevant and it picks different ones than Biznar does. This makes sense given that Google has potentially many more sources and search results to work with for any given query.
The question you should ask yourself is whether the additional results from Biznar make it worth having two alert services. I think so but you have to decide for yourself.
If you enjoyed this post, make sure you subscribe to the RSS feed!
Tags: biznar, federated search
2 Responses so far to "Biznar alerts: followup"
April 6th, 2009 at 2:32 pm
In a world where Google gives you literally millions of pages including most search terms, most of the time including _every_ page that mentions my search term is probably not what I want.
So my question would be not which returns more, but which returns _better_. It would be interesting to somehow evaluate whether the extra hits in Biznar are actually useful in real world cases, or just noise that happened to mention the term. Of course, what is considered useful vs. noise is very context dependent, and very user dependent.
April 10th, 2009 at 4:44 am
Jonathan - I agree. I should have worded that second-to-last sentence to say “better” rather than “most.”