FOSDEM 2008, Drupal ApacheSolr module

Tue, 02/26/2008 - 01:51 — Robert Douglass

Here is the video of my ApacheSolr presentation at FOSDEM. I had some technical difficulties (thank you very much, Keynote) at the beginning, but you can fast-forward past that bit.

Comments

Tue, 02/26/2008 - 02:58 — JoG (not verified)

Put your presentation online?

Congratulations with your new job !

Will you put your presentation online e.g. on SlideShare like berkes, wmostrey and toemaz?

See -> http://www.slideshare.net/tag/fosdem2008

Thank you !

Wed, 02/27/2008 - 02:02 — John Willis (not verified)

Great Post

I just wrote a post the other day about why Aquia should do "enterprise search" with things like Hadoop and Solr.

http://www.johnmwillis.com/enterprise-search/go-big-or-go-home/

Also, I didn't catch to much info on how you loaded the data. Do you have any more information on that.

Great presentation.
Thanks
johnmwillis.com

Wed, 02/27/2008 - 12:42 — Robert Douglass

The data gets loaded on cron from the normal Drupal db

It uses essentially the same technique that core search uses. It finds new and changed node content, builds it, and sends the text representation (via POST) to the Solr server.

-Rob

Thu, 02/28/2008 - 01:19 — John Willis (not verified)

Sorry to be pesky...

Could you post an example or send me an example of the actual posted data.

Thanks
johnmwillis.com

john dot willis @ zabovo.com

Thu, 02/28/2008 - 22:25 — Ricky Leung (not verified)

solr demo not up

Interesting video. I tried running the demo with what you suggested in the beginning about google..

so I tried "drupal google search robert douglass"

result
-google now has it ranked 3rd on the first page.
-solr demo does not have it on the first page but it is first on the second page. Meaning it is ranked 11.

Assuming your database is pretty recent in comparison to d.o., I tried the same thing there too.

-1st result on the 5th page, or ranked 51. I wasn't going to go beyond the 5th page.

My conclusion..
Google is still better, but solr search is a huge improvement over the one d.o. is using.

Fri, 02/29/2008 - 00:50 — Robert Douglass

While it's important to keep in mind that Google's results aren't necessarily the _right_ answer, I do think they're a good yardstick. We can (and should) do development on Drupal's search engine (I'm preparing a personal D7 battle plan for Drupal search), and the Solr demo I have set up has a lot of room for tweaking and improvement. If we can get a lot of people looking at the code and trying things out, I'm pretty confident we'll be able to elevate the quality of Drupal search significantly.

-Rob

Sat, 06/13/2009 - 13:22 — Drupal Development (not verified)

Thanks for sharing

Thanks for sharing information.

RobsHouse.net