I think many of you have ever had to deal with screwing the search on the site. So I work, I had to do it.
In search of a search engine, so understood, the leading places are occupied by
Lucene and
Sphinx .
But the requirements are the following (norm):
- Stable work with the Russian language
- Morphology ("guitar" <-> "guitar" <-> "guitars")
- Speed ​​performance
- Ability to index PG, MySQL and files (.html)
- PHP-API, installation on any * nix distribution (FreeBSD, Ubuntu Server)
')
There was a case, put Sphinx, set up (sort of like O_o), indexed the base with the catalog of goods. (included stemm_enru morphology).
Sort by relevance, I press: “fender guitar”, and from the database, first of all, it draws fender strings for me, then the guitars themselves, but it should be the other way around. I do not understand what is the matter, and what kind of weight and how to customize them.
I read the manual on the developer’s site, but everything is described as too complicated, or else I’m a sucker)
In general, dear, help please understand.