Technorati just released to the world the completely rewritten keyword search system we've been working on, and we're soliciting feedback!
The major improvements of this release are:
- Ability to keep up with the ever-increasing number of new weblogs and weblog updates each day (over 11,000 new weblogs created every day, over 100,000 updates every day). Median time from post to live index is 7 minutes. Over 1.6 Million sources tracked. I know of no other engine that even comes close.
- Post-level search, which means we can add synthetic RSS/Atom/whatever feeds whenever we're ready
- Keyword highlighting of search terms
- Permalinks to each post (well, they *should* work)
- Some advanced search capabilities, including using double quotes to search phrases, + as the AND operator, and - as the NOT operator:
Full integration with the current Link Cosmos search - it uses a pretty smart algorithm to figure out if you're searching on a URL or not - and if it sees a URL, it will do a Link Cosmos search by default, otherwise it performs a keyword search. You can also force a search type using a link on the results page, in case we guess wrong.
Oh, and we've made some significant speedups on getting the cosmos of big sites, like the New York Times, CNN, Google, Yahoo, and bOingbOing. Give 'em a whirl. Amazon is my favorite, it is like getting a quick glimpse into people's buying habits at any given time of the day.
We're looking for your feedback! Some areas where we already know we need improvement: Response time is important, and our goal is to get to under one second search times. For now, most keyword searches run in under 10 seconds. There are occasional duplicate posts in the database, we're working on cleaning those out as well. We're also getting the search XML APIs ready to go, they aren't quite ready to launch today.
Please let us know what else we can do to improve!
The new search is up on the Technorati site, so go and give it a try, and by all means, let us know what you think.