[elasticsearch] compute K-nearest neighbor for training a classifier purposes

knn.sh

##########################################################################################
# use case: training a classifier
# 
# Many systems classify documents by assigning “tag” or “category” fields. Classifying
# documents can be a tedious manual process and so in this example we will train a classifier
# to automatically spot keywords in new documents that suggest a suitable category.

curl -XGET "http://localhost:9200/products_fr/_search" -d'
{
  "query": {
    "function_score": {
      "query": {
        "query_string": {
          "query": "samsung",
          "default_operator": "AND",
          "fields": [
            "title^3",
            "description"
          ]
        }
      },
      "functions": [
        {
          "script_score": {
            "script": "_score * log1p(doc[\"hits\"].value) + log1p(doc[\"hits\"].value)"
          }
        }
      ]
    }
  },
  "aggregations": {
    "knn": {
      "significant_terms": {
        "field": "title","size": 3
      }
    }
  }
  , "size": 0
}'

knn_simple_query.sh

##########################################################################################
# use case: training a classifier
# 
# Many systems classify documents by assigning “tag” or “category” fields. Classifying
# documents can be a tedious manual process and so in this example we will train a classifier
# to automatically spot keywords in new documents that suggest a suitable category.

curl -XGET "http://localhost:9200/products_fr/_search" -d'
{
  "query": {
    "query_string": {
      "fields": ["title","description"],
      "query": "galaxy"
    }
  },
  "aggregations": {
    "knn": {
      "significant_terms": {
        "field": "title","size": 3
      }
    }
  }
  , "size": 0
}'

Can you please explain how this can be used for classifying new documents. I have something similar to do, and this might work.

Nothing related to K-Nearest-Neighbour here. Provided queries are Aggregations. It is not even "More-Like-This" type.