Archive for January, 2005

It looks like Google is starting to track which search results people select. This technique is used in many enterprise search products to allow the administrator to help users refine searches, create synonym lists, etc.

They aren’t doing it for all searches but I found one query. If you search for ‘oyster bay wines’ on Google UK, the result links will go to a preprocessor rather than directly to the site. Check out the status bar in the image:

I would be surprised if Google has started to manually analyze their searches so I assume they now have some clever software that does it for them.

Yahoo! has had a separate search page for a while. Initially it was a copy of the simplicity of the Google home page (see Google’s reasons, although I’ve also heard it’s like that because the founders weren’t particularly good at HTML). I always thought that was a good idea because the main Yahoo! page has become so cluttered over the years it’s now annoying to use.

But it looks like they couldn’t stay away. The Yahoo! search page now features news headlines and some stock information. I wonder how long it will be until the two pages are identical.

Last week I had a programming problem, I had to do a query expansion in Lucene. This particular expansion was to convert term and phrase queries into span queries to provide higher relevancy scores for results when the search terms were closer together.

I simplified the problem by only considering terms and phrases and boolean combinations of them. Basically:

• TermQuery becomes SpanTermQuery
• PhraseQuery becomes SpanNearQuery
• All those result span queries are combined into a SpanNearQuery

BooleanQueries would be handled recursive following the above convention. A BooleanQuery is essentially a list of BooleanClauses, so at this point those functional programming classes from school popped in my head. I have lists, recursion, base cases, it should be straightforward to jot down a solution in Haskell then I’ll translate it to Java. After a while I gave up and wrote a long piece of procedural code in Java, I just didn’t know Haskell well enough to write out anything more than trivial code.

Read the rest of this entry »

This post kicks off my weblog. I’ve had web sites before, but at some point or another I would stop updating them, mostly because I’d have nothing new to post. I hope this web log will be different. So why am I doing this?

I need to get better at writing. In his last post, Joel reflects what I’ve found since I started working. If you can’t communicate well you lose credibility fast, no matter how clever you think you are. A weblog gives me the chance to write more often and in a more casual matter.

It gives me a place to post my thoughts on various things. But I will try to avoid posting the same news as all the other blogs I read, unless of course, I have something interesting to add.

Thirdly…I don’t have a third, not a good one anyone.

So here I go.