How to handle multi-word synonyms

Question

I'm trying to understand the results that I'm getting in Elastic search in a couple of conditions. I have this list of synonyms defined:

"product insert, product inserts, qc package, qc package inserts, qc package insert, package insert => package inserts"

My hope was that all of the terms to the left of the arrow would be treated as the term to the right. Here is my index setting:

PUT /test_index
{
  "settings": {
    "index": {
      "analysis": {
        "filter": {
          "my_syn_filt": {
            "tokenizer": "keyword",
            "type": "synonym",
            "synonyms": [
              "product insert, product inserts, package inserts, qc package, qc packages, qc insert, qc inserts, package insert, qc package insert, qc package inserts => package inserts"
            ]
          }
        },
        "analyzer": {
          "my_synonyms": {
            "filter": [
              "lowercase",
              "my_syn_filt"
            ],
            "tokenizer": "keyword"
          }
        }
      }
    }
  }
}

My problem is that when I search a couple of the terms - "product insert", I don't get the results I expect. But "product inserts" works just fine. Is there something wrong with my configuration? Am I missing a step?

Answer 1

I've tested your settings and my guess is that you haven't assign my_synonyms analyzer to your field.

Without knowing how you define your mappings i will show you a working example:

Assuming your mapping and settings look like:

PUT /my_index
{
  "mappings": {
    "properties": {
      "data": {
        "type": "text",
        "analyzer": "my_synonyms",  => my guess
        "fields": {
          "keyword": {
            "type": "keyword",
            "ignore_above": 256
          }
        }
      }
    }
  },
  "settings": {
    "index": {
      "analysis": {
        "filter": {
          "my_syn_filt": {
            "tokenizer": "keyword",
            "type": "synonym",
            "synonyms": [
              "product insert, product inserts, package inserts, qc package, qc packages, qc insert, qc inserts, package insert, qc package insert, qc package inserts => package inserts"
            ]
          }
        },
        "analyzer": {
          "my_synonyms": {
            "filter": [
              "lowercase",
              "my_syn_filt"
            ],
            "tokenizer": "keyword"
          }
        }
      }
    }
  }
}

Indexing some data:

POST my_index/_doc/1
{
  "data":"package inserts"
}

Query which utilities the synonyms:

GET my_index/_search
{
  "query": {
      "match": {
        "data": "product insert"
      }
  }
}

Results:

{
 "took" : 0,
  "timed_out" : false,
  "_shards" : {
    "total" : 1,
    "successful" : 1,
    "skipped" : 0,
    "failed" : 0
  },
  "hits" : {
    "total" : {
      "value" : 1,
      "relation" : "eq"
    },
    "max_score" : 0.2876821,
    "hits" : [
      {
        "_index" : "my_index",
        "_type" : "_doc",
        "_id" : "1",
        "_score" : 0.2876821,
        "_source" : {
          "data" : "package inserts"
        }
      }
    ]
  }
}

Without assigning the analyzer to your field you will get results only if one of the words - package or inserts are included in your search query,actually without the analyzer you executing a simple match query which used the default elasticsearch standard analyzer.

Hope this helps