Comments on: Web application step 5: implement a “relevant” search engine

By: tim

tim — Wed, 07 Jan 2009 22:46:48 +0000

The Yahoo BOSS team came to paris a couple months ago. http://ysearchblog.com/2008/11/17/boss-hack-world-tour-heads-to-europe/

It was a really interesting talk & convinced me that BOSS is the best tool for data aggregation on the web right now. Google doesn’t compete yet. (besides, Google custom search is fairly poor when it comes to features … Google, your turn now !)

By: Thibauld

Thibauld — Wed, 07 Jan 2009 22:23:20 +0000

Yahoo BOSS looks very interesting indeed, I had never heard about it before! Thanks for the info, I will definitely investigate on this.

By: tim

tim — Wed, 07 Jan 2009 21:49:14 +0000

You should give Yahoo BOSS a fair try. No easiest way to search through ALL of your online content (tweets, delicious bookmarks, sites …)

By: Jan Seidl

Jan Seidl — Tue, 06 Jan 2009 18:00:03 +0000

Thi,

I’ve worked with keyword index tables that worked pretty nice. You need to populate the index table on insert/update of the field that will be searched for.

Working with LIKE is totally fail, as it slows down and return unrelevant data. With index tables you can have density and thus relevancy!

It’s quite neat and easy to mantain, realtime and quick!

I’ve wrote about it @ my blog about effective data searching, take a look if interested.

[]s!

By: Thibauld

Thibauld — Mon, 05 Jan 2009 21:32:08 +0000

I did not know about Sphinx, thanks for the info. It looks like an attempt to bring MySQL fulltext search capabilities close to PostgreSQL’s No kidding, next time I have to deal with MySQL, I will definitely consider using it.

By: Menno

Menno — Mon, 05 Jan 2009 10:34:53 +0000

Nice writeup about using the full-text capabilities for search. For our custom search-engine we’ve used Sphinx as the backend, it is highly configurable and performs very well. You should definitely check that out as well, as there are many good interfaces for ruby and other languages available.