 |
 |
> I have been having fun with KinoSearch (an open source indexer/search > engine with a Perl API), and I have documented my experiences here: > > http://dewey.library.nd.edu/morgan/kinosearch/?cmd=about > > Cool! > > -- > Eric Lease Morgan I'm taking a break from evaluating search engines to respond to this. I'm glad Eric took a look at this SE, and I appreciate this documentation. But I have to take issue with his labeling spell-check and synonym support as "featuritis." These are fundamental capabilities not so much "increasingly desired by users" but expected functionalities for search. The fielded search issues intrigued me, not because users do a lot of fielded searching (they don't), but because it reminded me that Kino doesn't support any faceting, either. As Eric notes, he would need to build in a spell-checker (and the ability to build a dictionary from his index). He didn't bring up stemming or light pluralization, but that would be yet another important capability. Then there's the issue of weighting fields. Then there's search log reporting. I could go on... but a search engine capable of supporting modern functionalities doesn't equate to a product larded with "featuritis." And if you say "but it's free" I'll scream... I agree on the indexing and Unicode issues with swish-e, which we will be migrating from shortly for a number of reasons. I also like Eric's idea of attaching search software to a major project. A search engine that had the collective strength and wisdom of LAMP software could be an interesting alternative to... *you know who.* Sorry if I overreacted to the "featuritis" comment--I live and die by search these days... I've already had to explain to more than one stakeholder why we can't just use Jimbob's Crapola Indexer or whatever. K.G. Schneider [log in to unmask]