Fuzziness in span query losing 1 edit distance

Elastic Stack Elasticsearch
1

Hi,

I am trying to do a fuzzy matching with phrase search. (This is already mentioned here : Combine Phrases search AND fuzzy matching and here : Elasticsearch Fuzzy Phrases - Stack Overflow)
I managed to solve the issue using span queries as mentioned in the posts however I have an issue.

For ex; If I have this document

{
  "name" : "Gracious Knight"
}

a query like this searches correctly

GET [INDEX]/_search
{
	"query": {
		"span_near": {
			"clauses": [
				{
					"span_multi": {
						"match": {
							"fuzzy": {
								"name": {
									"fuzziness": "2",
									"value": "Gracious"
								}
							}
						}
					}
				},
				{
					"span_multi": {
						"match": {
							"fuzzy": {
								"name": {
									"fuzziness": "2",
									"value": "Knight"
								}
							}
						}
					}
				}
			],
			"slop": 0,
			"in_order": "true"
		}
	}
}

However the issue is the fuziness edit distance is losing 1 level. So in the "fuzzy" under first "span_multi", this works;

{
	"name": {
	"fuzziness": "2",
	"value": "Grxcious"
}

Meanwhile this doesn't;

{
	"name": {
	"fuzziness": "2",
	"value": "Grxxious"
}

So when with edit distance of 2, it's ok if 1 character is changed howevre if 2 characters are changed no records are returned. If fuziness is set to 0 even if you put in the exact phrase it doesn't match. So it seems like in this type of query 1 character already seems like changed but I couldn't figure out why?

Just wondering if anybody has any idea about this?

Thanks

2

Ok, found the answer for this. Lowercase tokenizer is applied to the values when indexing. And in the search queries I am giving the first letter as capital which is what eating up the edit distance. If you give the whole value as lowercase then everything works as expected.

Duh!

3

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.