One autosuggestion solution would be to query normal text fields. That way you have the whole feature set of Solr (like NGram filtered text for infix search or field queries to scope the search). If you also store the data of the text field you directly have the results to use as autosuggestions. Unfortunately in some applications (like ours) different documents often contain the same data for the same field. So this approach may return many duplications as search hits for exactly this field. If one could collapse the results by that field all duplications would be gone and you only distinct hits remain (what one would expect from autosuggestions). At least this is how I understood field collapsing.
Kai -- View this message in context: http://lucene.472066.n3.nabble.com/Different-options-for-autocomplete-autosuggestion-tp2678899p2701055.html Sent from the Solr - User mailing list archive at Nabble.com.
