CODEWITHSUNDEEP
elasticsearch
rubyprince
rubyprince

Reputation: 17803

Get matched keywords while searching on an analysed field

Is there a way to get only the matched keywords while searching on an analysed field. My case is I have a 'content' field (string analysed) against which a query is run like this:

GET /posts/post/_search?pretty=true
{
    "query": {
        "query_string": {
            "query": "content:(obama or hilary)"
        }
    },
    "fields": ["id", "interaction_id", "sentiment", "tweet_created_at", "content"]
}

I get output like this:

"hits": [
         {
            "_index": "posts_v1",
            "_type": "post",
            "_id": "51764639fdccca097f03d095",
            "_score": 2.024847,
            "fields": {
               "content": "UGANDA HILARY",
               "id": "51764639fdccca097f03d095",
               "sentiment": 0,
               "tweet_created_at": "2012-11-24T14:59:25Z",
               "interaction_id": "1e236478961ca480e0744001f05ca8b8"
            }
         },
         {
            "_index": "posts_v1",
            "_type": "post",
            "_id": "51c2bae26c8f1806cb000001",
            "_score": 1.9791828,
            "fields": {
               "content": "Obama in Berlin — looking back",
               "id": "51c2bae26c8f1806cb000001",
               "sentiment": 0,
               "tweet_created_at": "2013-06-20T08:18:39Z",
               "interaction_id": "1e2d98202c55a980e07493a024172cb6"
            }
         },
         {
            "_index": "posts_v1",
            "_type": "post",
            "_id": "51c3a6b06c8f185fcb000001",
            "_score": 1.7071226,
            "fields": {
               "content": "Knowing Barack Obama, Hilary Clintonr",
               "id": "51c3a6b06c8f185fcb000001",
               "sentiment": 0,
               "tweet_created_at": "2013-06-21T01:04:45Z",
               "interaction_id": "1e2da0e8fb5fa480e07407b3fa87ab72"
            }
         }
]

So, I need to have something like this:

"hits": [
         {
            "_index": "posts_v1",
            "_type": "post",
            "_id": "51764639fdccca097f03d095",
            "_score": 2.024847,
            "fields": {
               "content": "UGANDA HILARY",
               "id": "51764639fdccca097f03d095",
               "sentiment": 0,
               "tweet_created_at": "2012-11-24T14:59:25Z",
               "interaction_id": "1e236478961ca480e0744001f05ca8b8",
               "content_tags": ["hilary"]
            }
         },
         {
            "_index": "posts_v1",
            "_type": "post",
            "_id": "51c2bae26c8f1806cb000001",
            "_score": 1.9791828,
            "fields": {
               "content": "Obama in Berlin — looking back",
               "id": "51c2bae26c8f1806cb000001",
               "sentiment": 0,
               "tweet_created_at": "2013-06-20T08:18:39Z",
               "interaction_id": "1e2d98202c55a980e07493a024172cb6",
               "content_tags": ["obama"]
            }
         },
         {
            "_index": "posts_v1",
            "_type": "post",
            "_id": "51c3a6b06c8f185fcb000001",
            "_score": 1.7071226,
            "fields": {
               "content": "Knowing Barack Obama, Hilary Clintonr",
               "id": "51c3a6b06c8f185fcb000001",
               "sentiment": 0,
               "tweet_created_at": "2013-06-21T01:04:45Z",
               "interaction_id": "1e2da0e8fb5fa480e07407b3fa87ab72",
               "content_tags": ["obama", "hilary"]
            }
         }
]

Please note the content_tags field in the second hits structure. Is there a way to acheive this?

Upvotes: 0

Views: 408

Answers (1)

Nik
Nik

Reputation: 739

Elasticsearch doesn't support returning which terms matched which field directly though I think it could implement one reasonably easily as an additional "highlighter". I think you have two options at this point:

  1. Do something hacky with highlighting like asking for the text length to be the max(all_strings.map(strlen).max, min_highlight_length), strip the text that isn't highlighted, and dedupe. I believe min_highlight_length is 13 characters or something. That might only apply to the FVH, which I don't suggest you use, so maybe you can ignore that.

  2. Do two searches either via multisearch or sequentially.

Upvotes: 1

Related Questions