I seem to be stuck behind the logic of a Solr faceting-powered tag cloud. First of all, I\'m using OpenNLP to parse my docs and obtain relevant words out of it, so every single
I would suggest using a single words field that is multivalued and stores the list of words per document.
having unbound number of word\d+ fields will complicate things.
if you use a single words multivalued field you can get all the words along with their frequencies which should be enough for creating the tag cloud.