For those who use Wordpress as their blog software of choice, most are familiar with the default keyword search that comes with Wordpress. The use of this search is limited, although, because it uses an inflexible SQL database query to grab matching results.

The disadvantages of the default search become more apparent as a user’s blog grows in size. Here are a few:

1. Database full-text searches are slow, and get worse as the database grows
2. SQL LIKE queries return all matches, not necessarily relevant ones
3. Results are returned without a true hierarchy of relevancy
4. Relevancy cannot be tweaked

I’ve released a new search plugin named ‘wpSearch’ which leverages the Zend Framework Lucene library, a port of the popular open source Lucene search library written in Java. wpSearch uses Lucene to build a ’search index’ of all of a user’s posts before the plugin is installed, and continues to add documents as posts are added to the blog. wpSearch overrides the internal search mechanism of Wordpress by hooking into its core, and placing its own results into Wordpress.

The result is that Wordpress never knows the difference, and your blog works the same as usual — except for the fact that now you have a fast and relevant Google-like search on your blog.

The underlying library that wpSearch uses, Lucene, is a proven tool in implementing some high profile search solutions. To name a few:

• Digg
• Akamai
• Overture
• Technorati
• HotJobs
• Epiphany
• FedEx
• Mayo Clinic
• MIT
• New Scientist Magazine

LTech has used the .NET port of Lucene to implement an enterpise job search for EmployOn, with great success. The .Net port is currently in use by the social networking giant MySpace.

wpSearch overcomes the default Wordpress search limitations.:

1. Lucene was made for full-text searching. Queries typically take place in under 50 milliseconds.
2. wpSearch uses it’s own PHP StandardAnalyzer, a component that allows searching to be very flexible (searches like “managed” match words like “manage”, unlike in Wordpress).
3. Results are returned with a logical heirarchy, with matches in a title placed higher than matches in post content.
4. What is considered relevant can be tweaked.

But to stop there would be selling wpSearch short. It supports a number of advanced search features, courtesy of Lucene:

• Wildcard searches
• Boolean searches
• Stop word filtering (English)
• “Fuzzy” searches (sounds like …)

Currently, wpsearch 1.5 has been released in the Wordpress plugin repository. A little more about wpSearch can also be found on LTech’s wpSearch page, and also at my development blog CodeFury.

So far, wpsearch has been well received, with some users writing in-depth reviews on it. One particular review was written by Patrick Cushing at EnterVenture. He wrote a piece highlighting some of the core strengths of wpSearch, with some side-by-side result comparisons which highlight the features that a blog without wpSearch is missing. The full review is at the EnterVenture blog.

wpSearch is currently installed on LTech’s blog, so you can see it in action. Just use the search box in the side bar to check it out!

The wpSearch 1.5.1 will include comment searching, along with a few updates to keep wpsearch compatible accross all blogs. Until then, check out the latest version at http://wordpress.org/extend/plugins/wpsearch/ and let me know what you think: katzgrau@gmail.com.