sirdf.com

[Phoog]

Hope you guys are still using this forum...

Im building on a news searchengine right now and need my own ranking algorithms so Im thinking about buildning my own fulltext/inverted index database. What is you experience with that?

Should I use a existing system like oracle or mysql and create my inverted index in that or should I build the database from scratch? Also, Im using Perl as programminglanguage.

The reason I want to build my own database for that is beacuse Im using another language so Im going to use a custom stemming algorithm and so on...

Anyone in here running a newssearch today?

Have a nice day!
/j

[runarb]

If you can use a relational database like Oracle or MySQL depends on the number of pages you want to search.

Relational database isn’t really made for searching, but can handle some million of documents.

If you are planning to only index some tens of thousand of documents you can use a sql database. But if you are planning for more then 100 000 you probably should use an inverted index of some sort.

Have you looked on Plucene (http://search.cpan.org/dist/Plucene/lib/Plucene.pm) ? It is a Perl port of Lucene.