Ferret vs Lucene 2.3.1

ruby-talk
1

Has anyone run any comparisons between Ferret and Lucene 2.3.1?

I saw many comparisons between older versions claiming Ferret is 2x
faster but these are outdated and I cannot use these metrics to make
any sort of decisions. Lucene is claiming to have made major speed
improvements in version 2.3. While my company is looking into Lucene
2.3.1, I am trying to see if Ferret being faster and quite possibly
better than Lucene is a better direction (also I'd get a chance to
plug Ruby into the system :))

If anyone has done any metrics recently can you please post the
results you have, also please indicate how much time was spent
optimizing Lucene and Ferret, or only minimal optimizations were
attempted.

Thanks in advance to anyone who posts.

2

Which one gives you higher quality search results? Ferret is supposed to be a port of Lucene, but Lucene *is* Lucene. Better? Worse? The same? I don't know, but I'd rather be a bit slower and get more relevant matches and if the result set is different that would be my main decision point. I've tried acts_as_ferret and acts_as_solr in Rails apps and for the simple case where the data is pretty different, one record to the next, the results are pretty good. It's where you have clustering that a problem can occur. (Think about indexing topics about Ruby that contain words similar to "active".)

HTH

ยทยทยท

On Mar 14, 2008, at 7:24 AM, astonerbum@gmail.com wrote:

Has anyone run any comparisons between Ferret and Lucene 2.3.1?

I saw many comparisons between older versions claiming Ferret is 2x
faster but these are outdated and I cannot use these metrics to make
any sort of decisions. Lucene is claiming to have made major speed
improvements in version 2.3. While my company is looking into Lucene
2.3.1, I am trying to see if Ferret being faster and quite possibly
better than Lucene is a better direction (also I'd get a chance to
plug Ruby into the system :))

If anyone has done any metrics recently can you please post the
results you have, also please indicate how much time was spent
optimizing Lucene and Ferret, or only minimal optimizations were
attempted.

Thanks in advance to anyone who posts.