Re: query time customized boosting
Hi, Any inputs or experience that others have come across will be really helpful to know. Basically, its the same as page ranking but the information used to decide the rank is much more dynamic in nature.. Appreciate any inputs. Regards Monmohan On Wed, Apr 4, 2012 at 4:22 PM, monmohan wrote: > Hi, > My index is composed of documents with an "author" field. My system is a > users portal where they can have a friend relationship among each other. > When a user searches for documents, I would like to boost score of docs in > which author is friend of the user doing the search. Note that the list of > friends for a user can be potentially big and dynamic (changing as the user > makes more friends) > > Is there a way to do this kind of boosting at query time? I have looked at > External field, query elevator and function queries but it appears that > none > of them > > Since the list of friends for a user is dynamic and per user based, it > can't > really be added as a field in the index for each document so I am not > considering that option at all. > Regards > Monmohan > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/query-time-customized-boosting-tp3883743p3883743.html > Sent from the Solr - User mailing list archive at Nabble.com. >
Re: query time customized boosting
the problem is how do I determine for each document the degree of separation and then apply boosting for example - say there is a user A - with friends X, Y, Z and another User B with friends L, M if there is a doc in index D1, with author field as Z and another doc D2 in index with author as L, It will be difficult to store degree of separation per user as an index field on the doc D1 and D2.. because for user A D1 has higher relevance but for user B its D2.. Is there a way to write our own elevation component in Solr.. On Thu, Apr 5, 2012 at 12:29 PM, William Bell wrote: > If you have degree of separation (like friend). You could do something > like: > > ...defType=dismax&bq=degree_of_separation:1^100 > > Thanks. > > On Thu, Apr 5, 2012 at 12:55 AM, Monmohan Singh > wrote: > > Hi, > > Any inputs or experience that others have come across will be really > > helpful to know. > > Basically, its the same as page ranking but the information used to > decide > > the rank is much more dynamic in nature.. > > Appreciate any inputs. > > Regards > > Monmohan > > > > On Wed, Apr 4, 2012 at 4:22 PM, monmohan wrote: > > > >> Hi, > >> My index is composed of documents with an "author" field. My system is a > >> users portal where they can have a friend relationship among each other. > >> When a user searches for documents, I would like to boost score of docs > in > >> which author is friend of the user doing the search. Note that the > list of > >> friends for a user can be potentially big and dynamic (changing as the > user > >> makes more friends) > >> > >> Is there a way to do this kind of boosting at query time? I have looked > at > >> External field, query elevator and function queries but it appears that > >> none > >> of them > >> > >> Since the list of friends for a user is dynamic and per user based, it > >> can't > >> really be added as a field in the index for each document so I am not > >> considering that option at all. > >> Regards > >> Monmohan > >> > >> -- > >> View this message in context: > >> > http://lucene.472066.n3.nabble.com/query-time-customized-boosting-tp3883743p3883743.html > >> Sent from the Solr - User mailing list archive at Nabble.com. > >> > > > > -- > Bill Bell > billnb...@gmail.com > cell 720-256-8076 >
Re: TikaEntityProcessor in Solr1.4
typo: Also, is there a timeframe on Solr1. release? should be Also, is there a timeframe on Solr1.5 release? On Tue, Apr 27, 2010 at 8:10 AM, monmohan wrote: > > Hi, > I would like to use TikaEntityProcessor with Solr1.4. > https://issues.apache.org/jira/browse/SOLR-1358 shows that this is added > in > Solr1.5. Can anyone please point me on the steps to patch Solr 1.4 with > these changes (if this is possible/allowed). > Also, is there a timeframe on Solr1. release? > > Regards > Monmohan > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/TikaEntityProcessor-in-Solr1-4-tp758371p758371.html > Sent from the Solr - User mailing list archive at Nabble.com. >
Re: Connection Pool
Sorry for hijacking the thread, but I have an additional question Is there a way to achieve similar performance (SUSS like) when targeting extract request handler (/update/extract)? I guess one way can be to extract content on the client side and then use SUSS to send update request but then extraction needs to be taken care of locally in an asynchronous/batch manner. Regards Monmohan On Sun, May 16, 2010 at 5:19 AM, Lance Norskog wrote: > Connection spooling is specified by the underlying apache commons > connection manager when you create the Server. > > The SUSS does socket pooling by default and is the preferred way to do > concurrent indexing. There are some quirks in the Server > implementation set, and SUSS avoids them. Unless you are willing to > root around in the SolrJ Server code and understand exactly how it > works, stay with the SUSS. > > On Fri, May 14, 2010 at 6:44 AM, gabriele renzi wrote: > > On Fri, May 14, 2010 at 3:35 PM, Anderson vasconcelos > > wrote: > >> Hi > >> I wanna to know if has any connection pool client to manage the > connections > >> with solr. In my system, we have a lot of concurrency index request. I > cant > >> shared my connection, i need to create one per transaction. But if i > create > >> one per transaction, i think the performance will down. > >> > >> How you resolve this problem? > > > > The commonsHttpSolrServer class does connection pooling, and IIRC also > > the StreamingUpdateSolrServer. > > > > > > > > -- > > blog en: http://www.riffraff.info > > blog it: http://riffraff.blogsome.com > > > > > > -- > Lance Norskog > goks...@gmail.com >
