CODEWITHSUNDEEP
javaelasticsearch
amachado
amachado

Reputation: 1040

Elasticsearch completion suggestion case insensitive

I'm doing an auto-complete in java using elasticsearch, so for that I built and map with the following code:

{
   "settings": {
      "analysis": {
         "analyzer": {            
            "whitespace_analyzer": {
               "type": "custom",
               "tokenizer": "whitespace",
               "filter": [
                  "lowercase",
                  "asciifolding"
               ]
            }
         }
      }
   },
   "mappings": {
     "product": {  
        "properties": {
           "product_suggestions" : {
              "type" :            "completion",
              "index_analyzer" :  "whitespace_analyzer",
              "search_analyzer" : "whitespace_analyzer",

              "preserve_position_increments": false,
              "preserve_separators": false
            }

        }
     }      
   }
}

And put the following index:

{
    "product_suggestions" : {"input":["Apple II Lisa","Apple"]}
}

and

{    
    "product_suggestions" : {"input":["Iphone 5s","apple"]}
}

If in java I do:

SuggestRequestBuilder req = esClient.prepareSuggest("auction").addSuggestion(
        new CompletionSuggestionFuzzyBuilder("searchSuggestion").field("product_suggestions").text(query).size(10));

SuggestResponse suggestResponse = req.execute().actionGet();

With the query "app" I will get "apple","Apple" and "Apple II Lisa".

Someone knows how I can fix this?

Thanks

Upvotes: 3

Views: 1828

Answers (1)

amachado
amachado

Reputation: 1040

I just followed in other path... set lowercase all words before indexing and this way the elasticsearch will only output apple and "apple ii lisa" (the ones of my example). I notice that is what google does also!! This will also avoid more complicated operations and maybe speedup my app!!

Thank you to everyone who tried to help.

Upvotes: 3

Related Questions