Elastic Search nested multimatch query

2019-03-17 23:03发布

问题:

So my problem is basically the same as described here, however it still remains unanswered on the group.

My mapping:

{
    "abstract": {
        "properties": {
            "summary": {
                "type": "string"
            }
        }
    },
    "authors": {
        "type": "nested",
        "properties": {
            "first_name": {
                "type": "string"
            },
            "last_name": {
                 "type": "string"
            }
        }
    }
}

And I would like to perform a full-text search on both of these fields, probably unequally weighted. The query that comes to my mind, but unfortunately doesn't work, would be this:

{
    "query": {
        "bool": {
            "should": [{
                "multi_match": {
                    "query": "higgs boson",
                    "fields": ["abstract.summary^5", "author.last_name^2"]
                }
            }]
        }
    }
}

I don't get any results from the authors field, because of its nested mapping. I also can't get rid of the nested property - I use it for aggregations. Any elegant idea how to solve it?

回答1:

Changing your mapping to the following one which uses include_in_root: true will allow you to use the query you original wrote:

{
    "abstract": {
        "properties": {
            "summary": {
                "type": "string"
            }
        }
    },
    "authors": {
        "type": "nested",
        "include_in_root": true,
        "properties": {
            "first_name": {
                "type": "string"
            },
            "last_name": {
                 "type": "string"
            }
        }
    }
}

You may want to index inner objects both as nested fields and as flattened object fields. This can be achieved by setting include_in_parent to true. - Link

Note: include_in_root may be deprecated in future versions of elasticsearch in favor of copy_to.



回答2:

The only solution that I managed to work out, which is not handy nor elegant but somehow works is such query:

"query": {
    "bool": {
        "should": [
            {
                "nested": {
                    "path": "authors",
                    "query": {
                        "multi_match": {
                            "query": "higgs",
                            "fields": ["last_name^2"]
                        }
                    }
                } 
            },
            {
                "multi_match": {
                    "query": "higgs",
                    "fields": ["abstract.summary^5"]
                }
            }
        ]
    }
}

I'm also not sure if the boosting will work as expected, providing it's set in different queries. Any suggestions appreciated.