As Josh Tuberville of eHarmony explained in the intro to his talk on Day 1 of Lucene Revolution “Jazzed about Solr: People as a Search Problem” (what a great double entendre the title of this session turns out to be! ), Jazzed.com is a new offering from eHarmony, currently in the final stages of development.

Here are the slides for this session.

“Jazzed wasn’t simply built with Solr. It was built around Solr.” The implication: there wouldn’t—no, make that couldn’t—be a Jazzed if there were no Solr; search technology is at the heart of solving the problem the site aims to solve.

Unlike the focus of eHarmony.com on finding a soulmate to spend the rest of your life with, Jazzed is a “photo-dating” site, which involves a broader range of relationships and demands a whole different set of considerations applied to the people search problem. To meet these, Josh and his colleagues have spent a lot of time thinking about flexible rankings based on Boolean algebra, vector space analysis, and hybrids of these.

Most of the data they handle is pretty structured, revolving around variables like eye color, hair color, ethnicity, and religion. But they do factor in some text analysis as appropriate and required.

Josh offers some interesting insights into the kind of issues Jazzed has confronted including, for example, the various relationships that can happen with the ability to hide profiles from other users.

He also observes on the need to pay to attention not only to what people say, but also what they do, as inputs into the ranking problem. What happens if a user says she prefers blue-eyed blondes, but always initiates contact with brown-eyed brunettes?

At the end of the day, success in facilitating people search on Jazzed.com depends on testing queries and closely analyzing the results. And, of course, when users find themselves out on a great date, having the time of their life — that is the real test of search results.

Cross-posted with Lucene Revolution Blog. Tony Barreca is a guest blogger.This is one of a series of presentation summaries from the conference.