{ "analyzer": { "keyword": { "filter": [ "truncate_keyword" ], "tokenizer": "no_splitting", "type": "custom" }, "lowercase_keyword": { "filter": [ "truncate_keyword", "icu_normalizer", "icu_folding", "remove_empty" ], "tokenizer": "no_splitting", "type": "custom" }, "near_match": { "char_filter": [ "near_space_flattener" ], "filter": [ "truncate_keyword", "icu_normalizer" ], "tokenizer": "no_splitting", "type": "custom" }, "near_match_asciifolding": { "char_filter": [ "near_space_flattener" ], "filter": [ "truncate_keyword", "icu_normalizer", "icu_folding", "remove_empty" ], "tokenizer": "no_splitting", "type": "custom" }, "plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "prefix": { "char_filter": [ "near_space_flattener" ], "filter": [ "icu_normalizer" ], "tokenizer": "prefix", "type": "custom" }, "prefix_asciifolding": { "char_filter": [ "near_space_flattener" ], "filter": [ "icu_normalizer", "icu_folding", "remove_empty" ], "tokenizer": "prefix", "type": "custom" }, "short_text": { "filter": [ "icu_normalizer", "aggressive_splitting", "preserve_original_recorder", "icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "whitespace", "type": "custom" }, "short_text_search": { "filter": [ "icu_normalizer", "aggressive_splitting" ], "tokenizer": "whitespace", "type": "custom" }, "source_text_plain": { "char_filter": [ "word_break_helper_source_text" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "source_text_plain_search": { "char_filter": [ "word_break_helper_source_text" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "suggest": { "filter": [ "icu_normalizer", "suggest_shingle" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "suggest_reverse": { "filter": [ "icu_normalizer", "suggest_shingle", "reverse" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "invis_cleanup", "sudachi_combo_filter", "sudachi_char_map", "sudachi_tok_break" ], "filter": [ "sudachi_split", "sudachi_baseform", "sudachi_posfilter", "sudachi_ja_stop", "icu_normalizer", "icu_folding", "remove_empty", "sudachi_word_delim", "flatten_graph" ], "tokenizer": "sudachi_tok", "type": "custom" }, "text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "invis_cleanup", "sudachi_combo_filter", "sudachi_char_map", "sudachi_tok_break" ], "filter": [ "sudachi_split", "sudachi_baseform", "sudachi_posfilter", "sudachi_ja_stop", "icu_normalizer", "icu_folding", "remove_empty", "sudachi_word_delim", "flatten_graph" ], "tokenizer": "sudachi_tok", "type": "custom" }, "token_reverse": { "filter": [ "reverse" ], "tokenizer": "no_splitting", "type": "custom" }, "trigram": { "filter": [ "icu_normalizer" ], "tokenizer": "trigram", "type": "custom" }, "trigram_anchored": { "char_filter": [ "add_regex_start_end_anchors" ], "filter": [ "icu_normalizer" ], "tokenizer": "trigram", "type": "custom" }, "word_prefix": { "filter": [ "icu_normalizer", "prefix_ngram_filter" ], "tokenizer": "icu_tokenizer", "type": "custom" } }, "char_filter": { "arabic_extended_norm": { "type": "mapping", "mappings": [ "\\uFB8E=>\\u0643", "\\uFB8F=>\\u0643", "\\uFB90=>\\u0643", "\\uFB91=>\\u0643", "\\u06A9=>\\u0643", "\\u06AA=>\\u0643", "\\uFEDB=>\\u0643", "\\uFEDC=>\\u0643", "\\uFED9=>\\u0643", "\\uFEDA=>\\u0643", "\\uFBFC=>\\u064A", "\\uFBFD=>\\u064A", "\\uFBFE=>\\u064A", "\\uFBFF=>\\u064A", "\\u06CC=>\\u064A", "\\uFBE8=>\\u064A", "\\uFBE9=>\\u064A", "\\uFEEF=>\\u064A", "\\uFEF0=>\\u064A", "\\u0649=>\\u064A", "\\u06CD=>\\u064A", "\\uFBE4=>\\u064A", "\\uFBE5=>\\u064A", "\\uFBE6=>\\u064A", "\\uFBE7=>\\u064A", "\\u06D0=>\\u064A", "\\uFBA6=>\\u0647", "\\uFBA7=>\\u0647", "\\uFBA8=>\\u0647", "\\uFBA9=>\\u0647", "\\u06C1=>\\u0647", "\\u06C0=>\\u0647", "\\uFBA4=>\\u0647", "\\uFBA5=>\\u0647", "\\u06D5=>\\u0647" ] }, "dotted_I_fix": { "type": "mapping", "mappings": [ "\u0130=>I" ] }, "globo_norm": { "type": "mapping", "mappings": [ "`=>'", "\u00b4=>'", "\u02b9=>'", "\u02bb=>'", "\u02bc=>'", "\u02bd=>'", "\u02be=>'", "\u02bf=>'", "\u02cb=>'", "\u055a=>'", "\\u05F3=>'", "\u2018=>'", "\u2019=>'", "\u201b=>'", "\u2032=>'", "\u2035=>'", "\ua78c=>'", "\uff07=>'", "\uff40=>'", "\\u202F=>\\u0020", "\u02c8=>", "\u02cc=>", "\\u0640=>", "\u066c=>,", "\u060c=>,", "\u055b=>", "\u055c=>", "\u055e=>", "\u00b5=>\u03bc", "\\u05F0=>\\u05D5\\u05D5", "\\u05F1=>\\u05D5\\u05D9", "\\u05F2=>\\u05D9\\u05D9", "\\uFB1F=>\\u05D9\\u05D9\\u05B7", "\\u05D9\\u05B7\\u05D9=>\\u05D9\\u05D9\\u05B7", "\u0d23\u0d4d\\u200D=>\u0d7a", "\u0d28\u0d4d\\u200D=>\u0d7b", "\u0d30\u0d4d\\u200D=>\u0d7c", "\u0d32\u0d4d\\u200D=>\u0d7d", "\u0d33\u0d4d\\u200D=>\u0d7e", "\u09a4\u09cd\\u200D=>\u09ce", "\\u061C=>" ] }, "icu_normalizer": { "type": "icu_normalizer", "name": "nfkc_cf" }, "invis_cleanup": { "type": "mapping", "mappings": [ "\\u200B=>\\u0020", "\\u00AD=>", "\\u200C=>", "\\u200D=>", "\\u2060=>", "\\uFEFF=>", "\\u200E=>", "\\u200F=>", "\\u202A=>", "\\u202B=>", "\\u202C=>", "\\u202D=>", "\\u202E=>", "\\u2066=>", "\\u2067=>", "\\u2068=>", "\\u2069=>", "\\u2061=>", "\\u2062=>", "\\u2063=>", "\\u2064=>", "\ufe00=>", "\ufe01=>", "\ufe02=>", "\ufe03=>", "\ufe04=>", "\ufe05=>", "\ufe06=>", "\ufe07=>", "\ufe08=>", "\ufe09=>", "\ufe0a=>", "\ufe0b=>", "\ufe0c=>", "\ufe0d=>", "\ufe0e=>", "\ufe0f=>", "\udb40\udd00=>", "\udb40\udd01=>", "\udb40\udd02=>", "\udb40\udd03=>", "\udb40\udd04=>", "\udb40\udd05=>", "\udb40\udd06=>", "\udb40\udd07=>", "\udb40\udd08=>", "\udb40\udd09=>", "\udb40\udd0a=>", "\udb40\udd0b=>", "\udb40\udd0c=>", "\udb40\udd0d=>", "\udb40\udd0e=>", "\udb40\udd0f=>", "\udb40\udd10=>", "\udb40\udd11=>", "\udb40\udd12=>", "\udb40\udd13=>", "\udb40\udd14=>", "\udb40\udd15=>", "\udb40\udd16=>", "\udb40\udd17=>", "\udb40\udd18=>", "\udb40\udd19=>", "\udb40\udd1a=>", "\udb40\udd1b=>", "\udb40\udd1c=>", "\udb40\udd1d=>", "\udb40\udd1e=>", "\udb40\udd1f=>", "\udb40\udd20=>", "\udb40\udd21=>", "\udb40\udd22=>", "\udb40\udd23=>", "\udb40\udd24=>", "\udb40\udd25=>", "\udb40\udd26=>", "\udb40\udd27=>", "\udb40\udd28=>", "\udb40\udd29=>", "\udb40\udd2a=>", "\udb40\udd2b=>", "\udb40\udd2c=>", "\udb40\udd2d=>", "\udb40\udd2e=>", "\udb40\udd2f=>", "\udb40\udd30=>", "\udb40\udd31=>", "\udb40\udd32=>", "\udb40\udd33=>", "\udb40\udd34=>", "\udb40\udd35=>", "\udb40\udd36=>", "\udb40\udd37=>", "\udb40\udd38=>", "\udb40\udd39=>", "\udb40\udd3a=>", "\udb40\udd3b=>", "\udb40\udd3c=>", "\udb40\udd3d=>", "\udb40\udd3e=>", "\udb40\udd3f=>", "\udb40\udd40=>", "\udb40\udd41=>", "\udb40\udd42=>", "\udb40\udd43=>", "\udb40\udd44=>", "\udb40\udd45=>", "\udb40\udd46=>", "\udb40\udd47=>", "\udb40\udd48=>", "\udb40\udd49=>", "\udb40\udd4a=>", "\udb40\udd4b=>", "\udb40\udd4c=>", "\udb40\udd4d=>", "\udb40\udd4e=>", "\udb40\udd4f=>", "\udb40\udd50=>", "\udb40\udd51=>", "\udb40\udd52=>", "\udb40\udd53=>", "\udb40\udd54=>", "\udb40\udd55=>", "\udb40\udd56=>", "\udb40\udd57=>", "\udb40\udd58=>", "\udb40\udd59=>", "\udb40\udd5a=>", "\udb40\udd5b=>", "\udb40\udd5c=>", "\udb40\udd5d=>", "\udb40\udd5e=>", "\udb40\udd5f=>", "\udb40\udd60=>", "\udb40\udd61=>", "\udb40\udd62=>", "\udb40\udd63=>", "\udb40\udd64=>", "\udb40\udd65=>", "\udb40\udd66=>", "\udb40\udd67=>", "\udb40\udd68=>", "\udb40\udd69=>", "\udb40\udd6a=>", "\udb40\udd6b=>", "\udb40\udd6c=>", "\udb40\udd6d=>", "\udb40\udd6e=>", "\udb40\udd6f=>", "\udb40\udd70=>", "\udb40\udd71=>", "\udb40\udd72=>", "\udb40\udd73=>", "\udb40\udd74=>", "\udb40\udd75=>", "\udb40\udd76=>", "\udb40\udd77=>", "\udb40\udd78=>", "\udb40\udd79=>", "\udb40\udd7a=>", "\udb40\udd7b=>", "\udb40\udd7c=>", "\udb40\udd7d=>", "\udb40\udd7e=>", "\udb40\udd7f=>", "\udb40\udd80=>", "\udb40\udd81=>", "\udb40\udd82=>", "\udb40\udd83=>", "\udb40\udd84=>", "\udb40\udd85=>", "\udb40\udd86=>", "\udb40\udd87=>", "\udb40\udd88=>", "\udb40\udd89=>", "\udb40\udd8a=>", "\udb40\udd8b=>", "\udb40\udd8c=>", "\udb40\udd8d=>", "\udb40\udd8e=>", "\udb40\udd8f=>", "\udb40\udd90=>", "\udb40\udd91=>", "\udb40\udd92=>", "\udb40\udd93=>", "\udb40\udd94=>", "\udb40\udd95=>", "\udb40\udd96=>", "\udb40\udd97=>", "\udb40\udd98=>", "\udb40\udd99=>", "\udb40\udd9a=>", "\udb40\udd9b=>", "\udb40\udd9c=>", "\udb40\udd9d=>", "\udb40\udd9e=>", "\udb40\udd9f=>", "\udb40\udda0=>", "\udb40\udda1=>", "\udb40\udda2=>", "\udb40\udda3=>", "\udb40\udda4=>", "\udb40\udda5=>", "\udb40\udda6=>", "\udb40\udda7=>", "\udb40\udda8=>", "\udb40\udda9=>", "\udb40\uddaa=>", "\udb40\uddab=>", "\udb40\uddac=>", "\udb40\uddad=>", "\udb40\uddae=>", "\udb40\uddaf=>", "\udb40\uddb0=>", "\udb40\uddb1=>", "\udb40\uddb2=>", "\udb40\uddb3=>", "\udb40\uddb4=>", "\udb40\uddb5=>", "\udb40\uddb6=>", "\udb40\uddb7=>", "\udb40\uddb8=>", "\udb40\uddb9=>", "\udb40\uddba=>", "\udb40\uddbb=>", "\udb40\uddbc=>", "\udb40\uddbd=>", "\udb40\uddbe=>", "\udb40\uddbf=>", "\udb40\uddc0=>", "\udb40\uddc1=>", "\udb40\uddc2=>", "\udb40\uddc3=>", "\udb40\uddc4=>", "\udb40\uddc5=>", "\udb40\uddc6=>", "\udb40\uddc7=>", "\udb40\uddc8=>", "\udb40\uddc9=>", "\udb40\uddca=>", "\udb40\uddcb=>", "\udb40\uddcc=>", "\udb40\uddcd=>", "\udb40\uddce=>", "\udb40\uddcf=>", "\udb40\uddd0=>", "\udb40\uddd1=>", "\udb40\uddd2=>", "\udb40\uddd3=>", "\udb40\uddd4=>", "\udb40\uddd5=>", "\udb40\uddd6=>", "\udb40\uddd7=>", "\udb40\uddd8=>", "\udb40\uddd9=>", "\udb40\uddda=>", "\udb40\udddb=>", "\udb40\udddc=>", "\udb40\udddd=>", "\udb40\uddde=>", "\udb40\udddf=>", "\udb40\udde0=>", "\udb40\udde1=>", "\udb40\udde2=>", "\udb40\udde3=>", "\udb40\udde4=>", "\udb40\udde5=>", "\udb40\udde6=>", "\udb40\udde7=>", "\udb40\udde8=>", "\udb40\udde9=>", "\udb40\uddea=>", "\udb40\uddeb=>", "\udb40\uddec=>", "\udb40\udded=>", "\udb40\uddee=>", "\udb40\uddef=>" ] }, "near_space_flattener": { "type": "mapping", "mappings": [ "'=>\\u0020", "\\u2019=>\\u0020", "\\u02BC=>\\u0020", "_=>\\u0020", "-=>\\u0020" ] }, "nnbsp_norm": { "type": "mapping", "mappings": [ "\\u202F=>\\u0020" ] }, "regex_acronym_fixer": { "type": "pattern_replace", "pattern": "(?<=(?:^|\\P{L})\\p{L})[.\uff0e](\\p{L})(?=\\P{L}|$)", "replacement": "$1" }, "regex_camelCase": { "type": "pattern_replace", "pattern": "(\\p{Ll}[\\p{M}\\p{Cf}]*)([\\p{Lu}\\p{Lt}])", "replacement": "$1 $2" }, "sudachi_char_map": { "type": "mapping", "mappings": [ "\uff08=>\\u0020", "\uff09=>\\u0020", "\u301c=>\uff5e" ] }, "sudachi_combo_filter": { "type": "pattern_replace", "pattern": "[\\u0300-\\u0362]", "replacement": "" }, "sudachi_tok_break": { "type": "pattern_replace", "pattern": "([^\u3001\u3002\\s\u300c\u300d\u30fb]{8000})", "replacement": "$1 " }, "word_break_helper": { "type": "mapping", "mappings": [ "_=>\\u0020", ":=>\\u0020", ".=>\\u0020", "(=>\\u0020", ")=>\\u0020", "\uff0e=>\\u0020", "\uff3f=>\\u0020", "\uff1a=>\\u0020", "\u00b7=>\\u0020" ] }, "word_break_helper_source_text": { "type": "mapping", "mappings": [ "_=>\\u0020", ".=>\\u0020", "(=>\\u0020", ")=>\\u0020", ":=>\\u0020" ] } }, "filter": { "aggressive_splitting": { "type": "word_delimiter_graph", "stem_english_possessive": false, "preserve_original": false }, "asciifolding": { "type": "asciifolding", "preserve_original": false }, "asciifolding_preserve": { "type": "asciifolding", "preserve_original": true }, "icu_folding": { "type": "icu_folding", "unicodeSetFilter": "[^\u304c-\u30fe]" }, "icu_nfkc_normalization": { "type": "icu_normalizer", "name": "nfkc" }, "icu_normalizer": { "type": "icu_normalizer", "name": "nfkc_cf" }, "lowercase": { "type": "lowercase" }, "prefix_ngram_filter": { "type": "edgeNGram", "max_gram": 255 }, "remove_empty": { "type": "length", "min": 1 }, "sudachi_posfilter": { "type": "sudachi_part_of_speech", "stoptags": [ "\u63a5\u7d9a\u8a5e", "\u52a9\u52d5\u8a5e", "\u52a9\u8a5e", "\u52a9\u8a5e,\u683c\u52a9\u8a5e", "\u52a9\u8a5e,\u526f\u52a9\u8a5e", "\u52a9\u8a5e,\u4fc2\u52a9\u8a5e", "\u52a9\u8a5e,\u63a5\u7d9a\u52a9\u8a5e", "\u52a9\u8a5e,\u7d42\u52a9\u8a5e", "\u52a9\u8a5e,\u6e96\u4f53\u52a9\u8a5e", "\u88dc\u52a9\u8a18\u53f7,\u53e5\u70b9", "\u88dc\u52a9\u8a18\u53f7,\u8aad\u70b9", "\u88dc\u52a9\u8a18\u53f7,\u62ec\u5f27\u958b", "\u88dc\u52a9\u8a18\u53f7,\u62ec\u5f27\u9589", "\u88dc\u52a9\u8a18\u53f7,\uff21\uff21", "\u88dc\u52a9\u8a18\u53f7,\uff21\uff21,\u4e00\u822c", "\u88dc\u52a9\u8a18\u53f7,\uff21\uff21,\u9854\u6587\u5b57", "\u7a7a\u767d" ] }, "sudachi_word_delim": { "type": "word_delimiter_graph", "split_on_case_change": false, "stem_english_possessive": true }, "suggest_shingle": { "type": "shingle", "min_shingle_size": 2, "max_shingle_size": 3, "output_unigrams": true }, "truncate_keyword": { "type": "truncate_norm", "length": 5000 } }, "normalizer": { "keyword": { "type": "custom", "filter": [ "truncate_keyword" ] }, "lowercase_keyword": { "type": "custom", "filter": [ "truncate_keyword", "icu_normalizer" ] } }, "tokenizer": { "no_splitting": { "type": "keyword" }, "prefix": { "type": "edgeNGram", "max_gram": 255 }, "sudachi_tok": { "type": "sudachi_tokenizer", "split_mode": "B" }, "trigram": { "type": "nGram", "min_gram": 3, "max_gram": 3 } } }