By Ellen Fox

As regular readers of Action for Libraries know, BCR has been expanding and upgrading our web site. We've added a lot of information, gussied up the graphics and reorganized the entire site to make for a clearer layout. At the last meeting of the BCR Internet Committee, we decided it would be helpful to add a search engine to the web site so BCR members could search the entire site by keyword.

The first step was to research search engines. There are plenty out there, but since BCR (like most of our member libraries) is a not-for-profit organization, cost was an important factor. Since BCR uses a UNIX web server, we concentrated on indexes that work on UNIX-based Apache web servers. Several free indexing packages are available for UNIX machines (widely available free software is one of the great advantages of UNIX systems). Harvest and WAIS-based systems are popular, but can be difficult to install and are best suited for different types of data. WAIS, in particular, works very well for fielded data, as users of GPO Access database know. But BCR's web site is largely filled with ordinary HTML pages, not heavily-fielded data.

We finally decided on SWISH-E, which stands for Simple Web Indexing System for Humans - Enhanced. It was created by Kevin Hughes and further developed by the Library of the University of California, Berkeley, which now maintains the official SWISH-E site (sunsite.berkeley.edu/SWISH-E/). SWISH-E allows for basic Boolean operators, relevance ranking and some fielded searching -- searchers can limit by title or by META tag information, for example.

Installing UNIX software can take from five minutes to a couple of days, depending on how many changes need to be made to the source code and whether anyone has provided clear documentation. In BCR's case, the code compiled with no difficulties, and we had the program in the proper directory within minutes. We borrowed and modified a Perl script (also available on the SWISH-E web site) to create a form for entering search queries and displaying the results. After that, the main task was to perform a variety of sample searches and see if any old, inactive documents were still hiding on the server somewhere. This housecleaning became the most time-consuming part of the whole process.

The final result: an easy-to-use keyword search engine on BCR's web site that we hope you will find useful. Looking for something specific on our site? Give it a try!