Elastic Search Function_Score Query with Query_String

Question

I was doing search using elastic search using the code:

es.search(index="article-index", fields="url", body={
  "query": {
    "query_string": {
      "query": "keywordstr",
      "fields": [
        "text",
        "title",
        "tags",
        "domain"
      ]
    }
  }
})

Now I want to insert another parameter in the search scoring - "recencyboost".

I was told function_score should solve the problem

res = es.search(index="article-index", fields="url", body={
  "query": {
    "function_score": {
      "functions": {
        "DECAY_FUNCTION": {
          "recencyboost": {
            "origin": "0",
            "scale": "20"
          }
        }
      },
      "query": {
        {
          "query_string": {
            "query": keywordstr
          }
        }
      },
      "score_mode": "multiply"
    }
  }
})

It gives me error that dictionary {"query_string": {"query": keywordstr}} is not hashable.

1) How can I fix the error?

2) How can I change the decay function such that it give higher weight to higher recency boost?

score 2 · Accepted Answer · answered Mar 17 '14 at 00:03

You appear to have an extra query in your search (giving a total of three), which is giving you an unwanted top-level. You need to remove the top-level query and replace it with function_score as the top level key.

res = es.search(index="article-index", fields="url", body={"function_score": {
    "query": {
        { "query_string": {"query": keywordstr} }
    },
    "functions": {
        "DECAY_FUNCTION": {
            "recencyboost": {
                "origin": "0",
                "scale": "20"
            }
        }
    },
    "score_mode": "multiply"
})

Note: score_mode defaults to "multiply", as does the unused boost_mode, so it should be unnecessary to supply it.

score 1 · Answer 2 · answered Feb 16 '15 at 20:26

1

You cant use dictionary as a key in the dictionary. You are doing this in the following segment of the code:

"query": {
    {"query_string": {"query": keywordstr}}
},

Following should work fine

"query": {
    "query_string": {"query": keywordstr}
},

answered Feb 16 '15 at 20:26

Pankaj Garg

893
2
8
15

score -1 · Answer 3 · edited Sep 17 '15 at 04:41

use it like this

     query: {
        function_score: {
          query: {
            filtered: {
              query: {
                bool: {
                   must: [
                      {
                        query_string: {
                          query: shop_search,
                          fields: [ 'shop_name']
                        },
                        boost: 2.0
                      },
                      {
                        query_string: {
                          query: shop_search,
                          fields: [ 'shop_name']
                        },
                        boost: 3.0
                      }
                  ]
                }
            },
            filter: {
      //          { term: { search_city:  }}
            }
          },
          exp: {
            location: {  
               origin: { lat:  12.8748964,
                lon: 77.6413239
              },
              scale: "10000m",
              offset: "0m",
              decay: "0.5"
            }
          }
  //        score_mode: "sum"
        }

Elastic Search Function_Score Query with Query_String

3 Answers3