I was involved in setting up a knowledge base for our IT Service Centre - that is, a database of articles with 1st line solutions, and information about customers and applications etc.

We were using Remedy's Action Request System as our Call logging/workflow tool, and had been using its internal Knowledge Base, which is very limiting - poor search facilities, limit on size of text, no formatting, attached files or hyperlinks. Plus, it's extremely slow.

We had a plan to buy an industry standard "knowledge solution" from somewhere... but these things take time... and we needed the solution now

So I wrote a Perl script to turn a report from Remedy into a set of HTML formatted documents. Then I wrote "my first Perl CGI script" to allow the user to search the documents, vote on their usefulness, and show which have been most recently accessed by other Service Centre users.

It's probably pretty noddy stuff - because I didn't think we'd have enough clout to get access to a database, I knocked up something based on the perl.com feature about creating a search engine using Search::Dict and plain text files. It works pretty fast for the number of documents we have (about 2000) but I'm sure it won't be scalable, and of course the data format needs serious munging if you want to do anything else with it. Also the search facility doesn't do boolean searches, and though the ranking of documents tends to give the best documents first, it will also often return a list of 100s of others...

But it works, and more importantly it gets used. Also, because we had a really good manager working on it, it spawned a whole set of processes to discuss and improve the solutions that go into it. Good work Perl!

So... a nice project might be to create a Knowledge Base system with:

• A real database (DBI::mysql ?)
• Boolean searching
• Better filtering of returned solutions
• Creating / editing / approving solutions online
;