If you use the suggested solution, it will detect the words at indexing time. However, Solr's FilterFactory's lifecycle keeps no track on whether a file for synonyms, keywords etc. has been changed since Solr's last startup. Therefore a change within these files is not visible until you reload your core.
Furthermore keywords for old documents aren't added automatically if you change your keywords (and reload the core) - you have to write a routine that finds documents matching the new keywords and reindex those documents. Example: Your keywordslist at time t1 contains two words: keyword codeword You are indexing two documents: doc1: {"content":"I am about a secret codeword."} doc1: {"content":"Happy keyword and the gang."} Your filter will mark "codeword" in doc1 and "keyword" in doc2 as words to keep and remove everything else. Therefore their content for your keepWordField contains only doc1: {"indexedContent":"codeword"} doc2: {"indexedContent":"keyword"} However, if you add the word "gang" to your keywordlist AND reload your SolrCore, doc2 will still only contain the term "keyword" until it gets reindexed again. Kind regards, Em Am 22.02.2012 17:56, schrieb Xavier: > I'm not sure to understand your solution ? > > When (and how) will be the 'word' detection in the fulltext ? before (by my > own) or during (with) solr indexation ? > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/How-to-merge-an-autofacet-with-a-predefined-facet-tp3763988p3767059.html > Sent from the Solr - User mailing list archive at Nabble.com. >