An idea of mine was to set up two SolrCores, since I don't know how to create a layer above popularity. Let's call them QueryCore and DataCore.
Querying for the first time against the DataCore would lead to a normal response, since there were no queries before. Let's say we are querying for "Star Wars". If the user is clicking on the link to document 3, this would lead to an update of QueryCore. QueryCore has got several fields: - query field - popularity field (popularity of the query) - query ID (stored) According to those SolrCores there is a document orientated database (DOD). The database's primary key is our query ID. It stores the first 10 responsed document IDs of our first query together with the clickthrough-rate per document for this query. If someone else is searching for "Star Wars" again, we will have a look at QueryCore and get the query ID. We are querying for this query-ID in our DOD and we are sorting the DataCore's response by the responsed clickthrough-rate. If one of the 10 documents of our first query will be deleted in future or is no longer at the top 10 responsed documents, we can update our DOD's data to keep it up-to-date. Querying the QueryCore+DOD and the DataCore could be done in two different threads. I emphasize: this is only an idea. I don't know how to do so only using Solr. Has anyone got other opinions or thoughts? -- View this message in context: http://old.nabble.com/Term-Dictionary-%2B-scoring-tp27174862p27197680.html Sent from the Solr - User mailing list archive at Nabble.com.
