Mercury Phrase classification

phrase-classification.json

{
  "settings": {
    "index": {
      "analysis": {
        "analyzer" : {
            "akt_analyzer_pattern_en" : {
                "tokenizer" : "akt_pattern_tokernizer_en",
                "filter" : [
                    "synonym_en",
                    "spell_en_US",
                    "lowercase",
                    "asciifolding",
                    "trim"
                ]
            },
            "akt_potential_type": {
                "tokenizer" : "akt_potential_type",
                "filter" : [
                    "lowercase",
                    "spell_en_US",
                    "asciifolding",
                    "trim"
                ]
            },
            "akt_text_without_type": {
                "tokenizer" : "akt_text_without_type",
                "filter" : [
                    "stop",
                    "lowercase",
                    "spell_en_US",
                    "asciifolding",
                    "trim"
                ]
            }
        },
        "tokenizer": {
          "akt_pattern_tokernizer_en": {
            "type": "pattern",
            "pattern": "(hotel|hostel|astotel|motel|restaurant|bar|museum|cafe|café|coffee|bakery|baker|pizzeria|pizza|^bake|bake$|cake|cuisine|food|bridge|stadium|moutain|beach|park|chocolat|salon|kitchen|boutique|resort|villa|bakery|cathedral|church|bank|airport|garden|spa$|^spa|^opera|opera$|station|studio|dessert|cuisine|cook|grill|steak| |,|-|_|&)"
          },
          "akt_potential_type": {
            "type": "pattern",
            "pattern": "(hotel[s]?|restaurant|bar|museum|cafe)",
            "flags": "CASE_INSENSITIVE",
            "group": 1
          },
          "akt_text_without_type": {
              "type": "pattern",
              "pattern": "(hotel[s]?|restaurant|bar|museum|cafe| |,|-|_|&)",
              "flags": "CASE_INSENSITIVE"
          }
        },
        "filter" : {
            "synonym_en" : {
                "type" : "synonym",
                "synonyms_path" : "analysis/en_synonyms.txt"
            },
            "spell_en_US": {
                "type":     "hunspell",
                "language": "en_US" 
            }
        }
      }
    }
  },
  "mappings": {
    "_doc": {
      "properties": {
        "object": {
          "type": "text",
          "analyzer": "akt_analyzer_pattern_en",
          "fields": {
            "name_without_type": { 
                "type": "text",
                "analyzer": "akt_text_without_type",
                "fielddata": true,
                "fields": {
                    "length": { 
                      "type":     "token_count",
                      "analyzer": "akt_text_without_type"
                    }
                }
            },
            "potential_type": { 
                "type": "text",
                "analyzer": "akt_potential_type",
                "fielddata": true
            }
          }
        },
        "type": {
          "type": "text"
        }
      }
    }
  }
}