3

I am examining AWS CloudSearch for system's new searching engine.

Assume that there are articles and some comments written on each articles. The search API should return articles that are matching or having any matching comments. So is there any possible way to retrieve DISTINCT values(in this case, unique ID of the article) from CloudSearch with single query execution? If not, what would be the nice solution to resolve this feature's requirement with CloudSearch?

I know there's text-array type for document field in CloudSearch but it seems expensive to update documents since N of comments on single article can be more than thousands.

dg99
  • 5,456
  • 3
  • 37
  • 49
wsn
  • 1,097
  • 8
  • 15
  • 1
    Rather than indexing an article with all its comments as an array, you could index a comment with the articleId it appeared in, and make the articleId a facet. Then when you search for a comment, your distinct articles will be that facet. – alexroussos Aug 04 '14 at 04:00

1 Answers1

1

I faced similar problem, putting comments is not an option in your case as array elements cannot be more than 1000 in cloudsearch. I will make two search domains, articles and comments. I will issue search query to both of them in parallel (async or multithreaded depending upon the language), articles will always generate non duplicate ids but on the results of comments query you have to apply the logic to an article id only once and always pick the top one, as results are sorted by matching score.

Muhammad Manazar
  • 158
  • 8