1. In order to test that I created a small dataset with 15 products of 3 brands.
PUT /test
PUT /test/products/1
{
"name" : "product 1",
"brand" : "brand1"
}
PUT /test/products/2
{
"name" : "product 2",
"brand" : "brand1"
}
PUT /test/products/3
{
"name" : "product 3",
"brand" : "brand1"
}
PUT /test/products/4
{
"name" : "product 4",
"brand" : "brand1"
}
PUT /test/products/5
{
"name" : "product 5",
"brand" : "brand1"
}
PUT /test/products/6
{
"name" : "product 6",
"brand" : "brand2"
}
PUT /test/products/7
{
"name" : "product 7",
"brand" : "brand2"
}
PUT /test/products/8
{
"name" : "product 8",
"brand" : "brand2"
}
PUT /test/products/9
{
"name" : "product 9",
"brand" : "brand2"
}
PUT /test/products/10
{
"name" : "product 10",
"brand" : "brand2"
}
PUT /test/products/11
{
"name" : "product 11",
"brand" : "brand3"
}
PUT /test/products/12
{
"name" : "product 12",
"brand" : "brand3"
}
PUT /test/products/13
{
"name" : "product 13",
"brand" : "brand3"
}
PUT /test/products/14
{
"name" : "product 14",
"brand" : "brand3"
}
PUT /test/products/15
{
"name" : "product 15",
"brand" : "brand3"
}
2. I did three queries: (A) one without any sorting, (B) a sort script using the Java hashCode function and (C) the last one using Elasticsearch function of random_score.
POST /test/products/_search
{
"from": 0,
"size": 3,
"query": {
"match": {
"name": "product"
}
}
}
POST /test/products/_search
{
"from": 0,
"size": 3,
"query": {
"match": {
"name": "product"
}
},
"sort": {
"_script": {
"script": "(doc['_id'].value + seed).hashCode()",
"type": "number",
"params": {
"seed": "doc['name'].value"
},
"order": "asc"
}
}
}
POST /test/products/_search
{
"from": 0,
"size": 3,
"query": {
"function_score": {
"query": {
"match": {
"name": "product"
},
"functions": [
{
"random_score": {
"seed": "1"
}
}
],
"score_mode": "sum"
}
}
}
}
3. The results were:
A. brand 2, brand 2, brand 1
B. brand 1, brand 3, brand 2
C. brand 3, brand 3, brand 1
4. My conclusion is that using the Java hash code function is the best approach. The ramdon_score function is interesting if you want to keep the results consistents for a same user (you can use the user id as the seed of this function).
Best regards,
Luiz
