Full text searching with SOLR
April 17, 2007
On the SOLR list there was a point brought up here about “progressive relaxation”. It’s a technique Oracle are promoting in their search product, but one of the SOLR devs dismisses it as flawed. The crux of the idea is for a developer to provide variation of queries, perhaps in succession, which represent more ‘open’ versions of the search term(s), leading to a possibly wider set of less accurate matches. The notion of SOLR’s ranking method seems to address this already, but I’m very new to that whole aspect of SOLR tuning, so I’m not sure who’s right here (perhaps both parties are?). Do you use developer-driven techniques for progressive search results, or do you just leave that up to the search engine you’re dealing with, or do you try to mix the two? I’m leaning towards the latter approach (mixing) but would appreciate any feedback.
Did you like this post? Buy me a hot chocolate!
Posted in Search, Software, Web, opensource
