Hello, >hi >we already have autosuggest working using solr based on popular search >terms.
Just terms of whole queries? I assume the latter. >we use following approach.. >http://www.lucidimagination.com/blog/2009/09/08/auto-suggest-from-popular-queries-using-edgengrams/ > >Now we want to use data indexed in solr also for autosuggest. with popular >search terms to have higher priority. > >can we just copy field containing doc text to a auto suggest filed which >does edgengram analysis? Something doesn't feel right here. Using data from the index for suggestions makes sense - we do that on http://search-lucene.com/ for example. Popular search terms having high priority and doc text, how does that work? Oh, you mean if you have a doc with field body whose value is "foo bar baz...." then, assuming the term "bar" is one of those popular search terms you would want "bar" to come up as a suggestion? That's doable with some coding, yes, but I don't think this would create a very good search experience. Here are some thoughts: * instead of suggesting popular query terms, suggest popular query strings * suggest phrases such as query strings, titles from a title field if you have it, author names from an author name field if you have it, and other fields of that nature * ... >also we have around 100 K docs in index so performance would be be a >concern? I think that depends on the implementation. For example, suggestions you see on search-lucene.com are powered by http://sematext.com/products/autocomplete/index.html and that solution works well with millions of suggestions. Otis ---- Sematext :: http://sematext.com/ :: Solr - Lucene - Nutch Lucene ecosystem search :: http://search-lucene.com/
