building a temporary index would certainly work, but it's a question of how efficient it would be (ie: how many users do you have, how often do they log in, how long does it take to build a typical index, how many concurrent users will you have, etc...)
one solution i've seen to a problem like this was to have a custom SearchComponent that did an external lookup to get the list of freinds, then for each friend did a search to get the DocSet of all their documents (letting them get cached in the filterCache) and computed the union of all those DocSets, and added that union as a filter for use by the QueryComponent. not sure if that type of approach will scale well to the number of users you are dealing with. : Date: Fri, 24 Apr 2009 02:22:37 -0700 (PDT) : Subject: Indexing tempary data on lucene : : : I have a list of public profiles of my site user's on solr index. There is : also a community around them, which is currently not their in Index. : : While searching, I have to give an option to search only my community : (friends and friends of friends). I could do it from data base query or : storing connection graph in memory but here I loose power of Solr Analyzers, : tokenizers and filters. : : Alternatively, I am thinking to store this relation temp in some other Solr : instance (running on a separate machine) and use it for search. I.e create : this index async when user logs in and destroy when user logs out. : : So when user searches for a profile the application will merge the results : from two indexes and returns unique users. : : Is this a practical/scalable solution? If yes, what performance : consideration, I should look for this new solr instance? For merging should : I built an application over solr or solr provides any way of merging results : from multiple indexes? : : Thanks, : Amit : : -- : View this message in context: http://www.nabble.com/Indexing-tempary-data-on-lucene-tp23212838p23212838.html : Sent from the Solr - User mailing list archive at Nabble.com. : -Hoss