Phil Bradley's Blog

Chat to me live if the image is green you can chat
with me if i'm online


Useful sites

Librarianship sites
Internet Resource Newsletter
Internet Scout
Librarian Avengers
Peter Scott's library blog
Shifted Librarian

Search engines

Search engine optimisation

Other useful sites
My personal weblog
Capital Radio
London Underground Map

Back to my home page


This page is powered by Blogger. Isn't yours?

RSS feed page

Monday, November 07, 2005

Searching the Guardian. Friend and fellow information professional Ian Winship alerted me to this story in the Guardian written by Emily Bell, editor in chief. The take on the article is that the newspaper has now made it easier to search their site. No more "baroque syntax" that delivered poor results, apparently.

All well and good so far; I'm certainly in favour of making life easier for people. However (you knew that was coming didn't you?) I think they've made a total dogs dinner of the thing. If you follow the link, you get taken to the page where this story is discussed, and there's a little 'Search this site' box to the left. It's very small, so you can't see more than 11 or 12 characters in your search string, which is annoying. And there's a 'Go' button. That's it. It is only when I've actually hit the Go button and the search has been run that I'm presented with some filters, such as sections (with a useful indication of the # of hits it's true), and the option of choosing publication year. If I already know that I just want stories from the weekly supplements for 2003 I have to run the search to begin with. It would be really nice if I could actually type all that into the search box to begin with, but apparently this would be some of that 'baroque syntax' that they're trying to get rid of.

It also gets worse. No help screen means that I have to guess what I can do. Two words seems to do an AND search and the results are based on relevant results. However, doing a phil bradley search gives me results that do contain both words, but they are separated in the results that I looked at. The first reference to Phil Bradley with both words next to each other is result #24. I fail to see how that improves relevance.

It's true that I can do a phrase search with "..." but I have to guess that. But wait, it gets worse! Suppose I want to find references to Microsoft, but without reference to Google. Microsoft NOT google doesn't work. Neither does microsoft not google do anything. Microsoft -google doesn't do anything at all either. So I'm making the assumption (and I could be wrong) that there is no way that I can do a search that excludes other words. This is better searching? There are over 600,000 articles in their database and there isn't even a NOT option?

Thanks, but I think I'll stick to baroque syntax if it's all the same to you.

Comments: Post a Comment