Elasticsearch Nested search with match - elasticsearch

Question

I've got orders index, and It has nested structure like below;
{
'baskets' => [
{
'id' => 123,
'product' => {
'title': 'blabla',
'tags': [
'value': 'Tag1',
'value': 'Tag2'
]
}
},
{
'id' => 1234,
'product' => {
'title': 'blabla2',
'tags': [
'value': 'Tag3',
'value': 'Tag2'
]
}
}
]
}
I want to make "similarity search" inside that informations like;
"Match inside tags, most match is the most closest. (For example If I search with Tag1, Tag2 It must be matched better"
But how can I make matching inside arrays in nested searchs?
I need to search like: where baskets.product.tags.value in ['Tag1', 'Tag2', 'Tag4']
Better matching must be ordered first.
My search query is below. But not working well.
['bool' =>
[
'should' => [
'nested' => [
'path' => 'baskets.product.tags',
'query' => ['terms' => ['baskets.product.tags' => $concept_array]]
]
],
]
]
]

Answer 1

Arrays in ES are not exactly treated as arrays, more like multiple objects in the same field so you can't easily score on multiple matches. Maybe with a score script but it will be very brittle.
Instead you can add a new field to the product, like all_tags, type text (you might need to change the separator and tune analyzers if your tags can contain spaces) and put all your tags in there, it will be "Tag1 Tag2" and "Tag3 Tag2" in your example. Then a simple
"match": {"baskets.product.all_tags": "Tag3 Tag2"}
inside your nested query should do the trick, scoring higher docs with closely matching baskets.