Edge_ngram explanation - elasticsearch

Question

Explain please what happens in followed case:
Index creation
curl --insecure -H "Authorization: ApiKey $ESAPIKEY" -X PUT "https://localhost:9200/tstind?pretty" -H 'Content-Type: application/json' -d'
{
"settings": {
"analysis": {
"filter": {
"cna_edge_ngram":
{ "type": "edge_ngram",
"min_gram": 3,
"max_gram": 10
}
},
"analyzer": {
"cna": {
"type": "custom",
"tokenizer": "whitespace",
"filter": ["lowercase", "cna_edge_ngram"]
}
}
}
},
"mappings": {
"properties": {
"cn": {
"type": "text",
"analyzer": "cna",
"fielddata": "true"
}
}
}
}
'
Test data
curl --insecure -H "Authorization: ApiKey $ESAPIKEY" -X POST "https://localhost:9200/tstind/_bulk?pretty" -H 'Content-Type: application/json' -d'
{"index": {"_index": "tstind", "_id": "1"}}
{"cn": "carrot"}
{"index": {"_index": "tstind", "_id": "2"}}
{"cn": "apple banana"}
{"index": {"_index": "tstind", "_id": "3"}}
{"cn": "redapple apple orange"}
{"index": {"_index": "tstind", "_id": "4"}}
{"cn": "orange"}
{"index": {"_index": "tstind", "_id": "5"}}
{"cn": "apple"}
{"index": {"_index": "tstind", "_id": "6"}}
{"cn": "cucumber"}
'
Request
curl --insecure -H "Authorization: ApiKey $ESAPIKEY" -X GET "https://localhost:9200/tstind/_search?pretty" -H 'Content-Type: application/json' -d'
{
"query": {
"match": {
"cn": {"query": "appls"}
}
}
}
'
Result:
{
"took" : 0,
"timed_out" : false,
"_shards" : {
"total" : 1,
"successful" : 1,
"skipped" : 0,
"failed" : 0
},
"hits" : {
"total" : {
"value" : 3,
"relation" : "eq"
},
"max_score" : 1.246899,
"hits" : [
{
"_index" : "tstind",
"_id" : "5",
"_score" : 1.246899,
"_source" : {
"cn" : "apple"
}
},
{
"_index" : "tstind",
"_id" : "2",
"_score" : 1.1766877,
"_source" : {
"cn" : "apple banana"
}
},
{
"_index" : "tstind",
"_id" : "3",
"_score" : 1.113962,
"_source" : {
"cn" : "redapple apple orange"
}
}
]
}
}
I'm expect that result must be empty with no hits. Why in results exists documents that doesn't contains requested phrase "appls"?
I'm trying to investigate with analyze, what tokens in index exists:
curl --insecure -H "Authorization: ApiKey $ESAPIKEY" -X GET "https://localhost:9200/tstind/_analyze?pretty" -H 'Content-Type: application/json' -d'
{
"analyzer": "cna",
"text" : "apple banana"
}
'
{
"tokens" : [
{
"token" : "app",
"start_offset" : 0,
"end_offset" : 5,
"type" : "word",
"position" : 0
},
{
"token" : "appl",
"start_offset" : 0,
"end_offset" : 5,
"type" : "word",
"position" : 0
},
{
"token" : "apple",
"start_offset" : 0,
"end_offset" : 5,
"type" : "word",
"position" : 0
},
{
"token" : "ban",
"start_offset" : 6,
"end_offset" : 12,
"type" : "word",
"position" : 1
},
{
"token" : "bana",
"start_offset" : 6,
"end_offset" : 12,
"type" : "word",
"position" : 1
},
{
"token" : "banan",
"start_offset" : 6,
"end_offset" : 12,
"type" : "word",
"position" : 1
},
{
"token" : "banana",
"start_offset" : 6,
"end_offset" : 12,
"type" : "word",
"position" : 1
}
]
}
Looks like is everything ok, but result not as expected. Explain please, what happened in this case?
I found the way how to filter results with post_filter, but I think it is not the best idea.

Answer 1

In this case look the token to term "appls".
{
"tokens": [
{
"token": "app",
"start_offset": 0,
"end_offset": 5,
"type": "word",
"position": 0
},
{
"token": "appl",
"start_offset": 0,
"end_offset": 5,
"type": "word",
"position": 0
},
{
"token": "appls",
"start_offset": 0,
"end_offset": 5,
"type": "word",
"position": 0
}
]
}
Now, token by term "apple":
{
"tokens": [
{
"token": "app",
"start_offset": 0,
"end_offset": 5,
"type": "word",
"position": 0
},
{
"token": "appl",
"start_offset": 0,
"end_offset": 5,
"type": "word",
"position": 0
},
{
"token": "apple",
"start_offset": 0,
"end_offset": 5,
"type": "word",
"position": 0
}
]
}
O match occur because token "app" exists in term "appls" and document "apple".