Basically, I got the same problem as this guy but I also need relevancy: PHP 'smart' search engine to search Mysql tables advice
But, I can't use Sphinx or Lucene (shared hosting)...
I can use LIKE but that has not relevancy so does anyone know a good PHP class that does create relevancy?
So I need something that fits this bill:
- Result relevancy value
- Match part of words, i.e. "LIKE %searchterm%"
- Search multiple databa开发者_JAVA技巧se columns
- Can't use things like Sphinx or Lucene, no supported by my shared hosting...
I'm using the CakePHP framework.
You might be interested in my presentation Practical Full Text Search in MySQL.
I compare performance of several solutions:
LIKE
predicates- MyISAM
FULLTEXT
indexes - Sphinx Search
- Lucene (or Solr)
- Google Custom Search Engine
- Inverted indexes
If you have no opportunity to install a dedicated search engine, I'd recommend using FULLTEXT or Google CSE.
Using LIKE
predicates is pretty clearly the worst choice.
I have tried Zend_Search_Lucene, the pure-PHP clone of Apache Lucene, but the PHP version is much too slow at indexing. See my answer to Can I predict how large my Zend Framework index will be? (and some quick Q:s)
Zend_Search_Lucene also hasn't maintained compatibility with new versions of the Apache Lucene index file format. See Using Solr and Zends Lucene port together…
Zend_Search_Lucene is an implementation of Lucene in pure PHP. I use it on a shared hosting website I run.
I ended up using mysql's full-text search in boolean mode. I also append the * operators to each search term.
user input: tom jones
resulting query: match(name,description) against('tom* jones* IN BOOLEAN MODE');
Thanks for everyone's help.
精彩评论