Elasticsearch: Boosting Query with negative_boost does not work as expected on 6.x

Created on 19 Feb 2018 · 5Comments · Source: elastic/elasticsearch

Elasticsearch version : 6.x

JVM version :1.8.0

OS version : Windows 10, Kubernetes Pods

Description of the problem:
On versions 6.x it seems that Boosting Query with negative_boost does not work anymore as described in the Official Guide.
On versions 5.x it works as expected.

Steps to reproduce:

1. create index mapping:

PUT /testn1

    {
    "mappings": {
        "test": {
            "dynamic": "strict",
            "_all": {
                "enabled": false
            },
            "properties": {
                "id": {
                    "type": "keyword"
                },
                "tags": {
                    "type": "keyword"
                }
            }
        }
    }
}

2. Insert 3 documents:

PUT /testn1/test/1

{
    "id": "1",
    "tags": ["active", "first"]
}

PUT /testn1/test/2

{
    "id": "1",
    "tags": ["active", "second"]
}

PUT /testn1/test/3

{
    "id": "1",
    "tags": ["active", "third"]
}

3. Execute a search using the following query:

/testn1/_search

{
    "query": {
        "boosting": {
            "positive": {
                "term": {
                    "tags": "active"
                }
            },
            "negative": {
                "term": {
                    "tags": "second"
                }
            },
            "negative_boost": 0.2
        }
    }
}

4. No matter what values we will use for the negative query, we will get the same result:

{
    "took": 2,
    "timed_out": false,
    "_shards": {
        "total": 5,
        "successful": 5,
        "skipped": 0,
        "failed": 0
    },
    "hits": {
        "total": 3,
        "max_score": 0.2876821,
        "hits": [{
                "_index": "testn2",
                "_type": "test",
                "_id": "2",
                "_score": 0.2876821,
                "_source": {
                    "id": "1",
                    "tags": [
                        "active",
                        "second"
                    ]
                }
            },
            {
                "_index": "testn2",
                "_type": "test",
                "_id": "1",
                "_score": 0.2876821,
                "_source": {
                    "id": "1",
                    "tags": [
                        "active",
                        "first"
                    ]
                }
            },
            {
                "_index": "testn2",
                "_type": "test",
                "_id": "3",
                "_score": 0.2876821,
                "_source": {
                    "id": "1",
                    "tags": [
                        "active",
                        "third"
                    ]
                }
            }
        ]
    }
}

I suspect that the cause may be in some code being changed in the Lucene library, BoostingQuery seems to be deprecated: https://issues.apache.org/jira/browse/LUCENE-8099. If this is the case, do we have an efficient alternative to demote results that match a given query ?

:AnalyticAggregations >bug

Source