Search for the Search Update

In March 2005, I discussed the Search for the Search My Place Of Work (MPOW) was embarking on. Since then, we went through the CMS upgrade (which was a gruesome experience until I fired the Vendor from Hell, and only slightly better thereafter, Hell being a very deep hole to fall into) and have focused on fixing Vendor From Hell’s legacy screw-ups so we could finish the deliverables first set out, oh, say, three years and a few wrinkles ago.

We are very happy with Nice Vendor, and though there are things I’d like to fix–the 1990s-ish HTML coding on our site is part of that–I am turning toward this year’s big deliverable–a new search engine.

I had debated writing about that on this blog, and understand I’ll walk a careful line as I do. I’ll stay focused on the why and how. In a nutshell, the “why” has two sides to it. One goal is to improve the user search experience, with the first emphasis on your typical user typing one or two words over and over again, and the second emphasis on the MPOW super-user who comes to MPOW with task knowledge of typical SE conventions and bells and whistles. Another is to stabilize the cost of managing and administering the search engine (henceforth, SE).

Our angelic new developers are very LAMP-savvy, but search naive. (“Can’t you just use Atomz?” Uh… no.) But when I say things like “Can you speak with Very Wise Search Engine Guru about how to generate search engine queries?” they perk up and wag their tails. They are very, very, very nice people. (Did you ever break up with someone and then meet the perfect someone? Ok, you know what I’m talking about. Even perfection is relative.)

What I’ve done so far:

* Reviewed MPOW’s budget–gulp, the Vendor From Hell ate a chunk of our SE development money; we can survive by shaking the couch cushions, though Expectations are ever so slightly lowered, and you can bet I’m on a tight leash with expenses overall

* Identified search engine products that met our minimum criteria and contacted said vendors, sometimes with cozy long walk-throughs

* Worked with a very savvy usability specialist and an equally sharp search engine guru to flesh out the activities required for SE evaluation

* Developed a timeline and GANNT chart for testing at least three SEs from March through May 31, with a product decision targeted for June 1 (My first GANNT chart! I do not even know what a GANNT is, but I’m still very proud of myself)

* Coordinated with MPOW’s Angelic Vendors to get estimates for the work required to install SE software and do whatever magical tweaks are required, plus ensure that our schedule didn’t conflict with their schedule

* Done enough research to help the Angelic Vendors identify how to generate the top spelling errors from our current spell-check, aspell (that’s for testing spell-check functions in SEs)

* With Savvy Search Engine Guru’s help, determined that we need to do search log analysis–SLA–in order to see what queries our users are using, and then determined that no, we do not produce these logs as we did on our old system, but yes, it’s feasible. Had a marvelous moment where I found a post about implementing search query logging on a website hosted on the same CMS we use, with examples that work almost out of the box. (One hopes.)

My next step–not a biggy, just a small prep while SLA logging gears up–is to take the SE criteria I’ve established and make a dedicated file for each product we’re evaluating.

Of the following topics, what would you like to hear about next?

* MPOW’s SE criteria–what, and why
* Why MPOW’s SE selection process is important for all you folks in LibraryLand with your pathetically unsearchable OPACs

Topics I’ll address as we go through the process:

* SLA: Why it’s important, and how to do it
* Spell-check error logging and analysis
* Search results and usability
* Text SEs versus Faceting SEs: so how does the user experience compare?