Function Score On Nested Object - elasticsearch

Question

I have this index blog with the following settings and mappings.
PUT /blog
{
"settings": {
"index": {
"number_of_shards": "1"
}
},
"mappings": {
"post": {
"_all": {
"enabled": false
},
"properties": {
"title": {
"type": "string"
},
"content": {
"type": "string"
},
"visitor": {
"type": "nested",
"properties": {
"id": {
"type": "string",
"index": "not_analyzed"
},
"last_visit": {
"type": "date",
"format": "yyyy-MM-dd"
}
}
}
}
}
}
}
I want to rank my posts based on relevancy and visitor's last visit. I tried this query without success. It seems like the gauss function cannot get the value of visitor's last_visit. How to get this worked?
POST /blog/post/_search
{
"query": {
"function_score": {
"functions": [
{
"gauss": {
"visitor.last_visit": {
"origin": "now/d",
"offset": "3d",
"scale": "4d",
"decay": 0.5
}
},
"filter": {
"nested": {
"path": "visitor",
"query": {
"term": {
"visitor.id": "1"
}
}
}
}
}
]
}
}
}

Answer 1

Here is a query with a match for a name that uses a nested object that I had for a particular use case. I didn't use any date fields, but as I said, it does use a nested object. I used relevancy of distance along with a text match, so it's similar.
I used the answer from this question to structure my query as it matched what I was trying to do. Scoring documents by text match and distance
GET dev_search_core_data/_search?size=200
{
"query": {
"bool": {
"should": [
{
"match": {
"NAME": "Amy Smith"
}
},
{
"bool": {
"must": [
{
"function_score": {
"query": {
"nested": {
"path": "LOCATION",
"query": {
"term": {
"LOCATION.SOME_IND": {
"value": true
}
}
}
}
},
"functions": [
{
"gauss": {
"LOCATION.COORDINATES": {
"origin": "-118.309, 34.041",
"scale": "50km",
"offset": "10km",
"decay": 0.5
}
}
}
]
}
}
]
}
}
]
}
}
}
I think the problem is with the structure of your query. I always run this command first to validate my queries if I'm having any problems to eliminate any syntax issues.
GET dev_search_core_data/_validate/query?explain
This was the result:
{
"valid": true,
"_shards": {
"total": 1,
"successful": 1,
"failed": 0
},
"explanations": [
{
"index": "dev_search_core_data_b",
"valid": true,
"explanation": "filtered((NAME:amy NAME:smith) (+function score (ToParentBlockJoinQuery (filtered(LOCATION.SOME_IND:true)->random_access(_type:_LOCATION)),function=org.elasticsearch.index.query.functionscore.DecayFunctionParser$GeoFieldDataScoreFunction#274227b9)))->cache(org.elasticsearch.index.search.nested.NonNestedDocsFilter#1012ada6)"
}
]
}
I also looked at the docs for an in-depth explanation of how the function score worked. You don't mention your version, but I'm using ES 1.6.