Elasticsearch Sorting Tiebreakers - elasticsearch

Question

Say I am creating a search engine for a photo sharing social network and the documents of the site have the following schema
{
"id": 123456
"name": "Foo",
"num_followers": 123456,
"num_photos": 123456
}
I would like my search results to satisfy the following requirements:
Only have results where the search query strings matches the "name" field in the document
Rank the search results by number of followers descending
In the case where multiple customers have the same number of followers, rank by number of photos descending
For example, say I have the following documents in my index:
{
"id": 1,
"name": "Customer",
"num_followers": 3,
"num_photos": 27
}
{
"id": 2,
"name": "Customer",
"num_followers": 25,
"num_photos": 1
}
{
"id": 3,
"name": "Customer",
"num_followers": 8,
"num_photos": 2
}
{
"id": 4,
"name": "Customer",
"num_followers": 8,
"num_photos": 5
}
{
"id": 5,
"name": "FooBar",
"num_followers": 10000,
"num_photos": 20000
}
If I search "Customer" in the search bar of the site, the ES hits should be in the following order:
{
"id": 2,
"name": "Customer",
"num_followers": 25,
"num_photos": 1
}
{
"id": 4,
"name": "Customer",
"num_followers": 8,
"num_photos": 5
}
{
"id": 3,
"name": "Customer",
"num_followers": 8,
"num_photos": 2
}
{
"id": 1,
"name": "Customer",
"num_followers": 3,
"num_photos": 27
}
I'm assuming I will need to perform some sort of compact query to create this "tiebreaker" logic. What clauses should I be using? If anyone had an example of something similar that would be amazing. Thanks in advance.

Answer 1

This sounds like a pretty standard sorting use case. Elasticsearch can sort on multiple fields in a predefined priority order. See documentation here.
GET /my_index/_search
{
"sort" : [
{ "num_followers" : {"order" : "desc"}},
{ "num_photos" : "desc" }
],
"query" : {
"term" : { "name" : "Customer" }
}
}
Obviously this is just a simple term query -- you may want that to be a keyword search instead based on the wording of your question.