Perhaps look into grouping on that field.
> On Mar 13, 2017, at 9:08 PM, Scott Smith <ssm...@mainstreamdata.com> wrote: > > I'm trying to solve a search problem and wondering if facets (or something > else) might solve the problem. > > Let's assume I have a bunch of documents (100 million+). Each document has a > category (keyword) assigned to it. A single document my only have one > category, but there may be multiple documents with the same category (1 to a > few hundred documents may be in any one category). There are several million > categories. > > Supposed I'm doing a search with a page size of 50. What I want to do is do > a search (e.g., "dog") and get back the top 50 documents that match the > contain the word "dog" and are all in different categories. So, there needs > to be one document from 50 different categories. > > If that's not possible, then is it possible to do it if I know the 50 > categories up-front and hand that off as part of the search (so "find 50 > documents that match the term 'dog' and there is one document from each of 50 > specified categories"). > > Is there a way to do this? > > I'm not extremely knowledgeable about facets, but thought that might be a > solution. But, it doesn't have to be facets. > > Thanks for any help > > Scott > >
