2

Looking for Open Source Search Federation solutions. Essentially I have a myriad of data repositories that I am unable to connect directly to SOLR.

I have evaluated SESAT (http://sesat.no/home.html) which is a possibility, however I am curious if there are other open source federated search solutions. Ideally I'd like to find something that can work along side SOLR or Lucene.

gombala
  • 83
  • 1
  • 6
  • +1 ... apache cocoon could be a (minor) option – Karussell Dec 28 '10 at 19:27
  • Cay you tell us more about which data sources are you using that are not supported by Solr? – Marko Bonaci Mar 08 '11 at 12:34
  • Federation to data that I might not be able to pull from SOLR. For example a Sybase database that we don't have the needed credentials to connect SOLR. Also, we will need to be able to connect to different GSA and FAST search engines that are hosted by different owners. – gombala Mar 09 '11 at 21:05
  • I don't get it, if you don't have needed credentials to access your Sybase data how are you planning to pull the data from any other IR solution? Any data source you can access through JDBC (com.sybase.jdbc2.jdbc.SybConnection for Sybase) is accessible by Solr's DataImportHandler. – Marko Bonaci Mar 09 '11 at 22:30
  • You federate to it's search interface... hence why I am asking for an open source search federation solution. I don't believe Solr can connect to a GSA and index the GSA's index (that wouldn't make any sense). Check out the wiki page on Federated Search if you are unclear - http://en.wikipedia.org/wiki/Federated_search – gombala Mar 10 '11 at 18:42
  • Got it, sorry for that, I mixed up federated and distributed. AFAIK, Solr doesn't have the option of federating other IR system outputs. In order to function properly Solr needs to have index/indices built, because, amongst other reasons, document scoring depends on index-time calculations, like Inverse document frequency, Term frequency, ... So, even if you could process results of another search engine on the fly, I don't see (with my modest IR knowledge though) how would you make searches relevant and fast at the same time. – Marko Bonaci Mar 10 '11 at 23:19
  • That would require [something like this](http://wunderwood.org/most_casual_observer/2007/04/progressive_reranking.html). – Marko Bonaci Mar 10 '11 at 23:23

0 Answers0