{ "analyzer": { "de_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper", "german_charfilter" ], "filter": [ "de_icu_normalizer", "preserve_original_recorder", "de_icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "de_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper", "german_charfilter" ], "filter": [ "de_icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "de_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "german_charfilter" ], "filter": [ "de_icu_normalizer", "german_stop", "german_normalization", "german_stemmer", "de_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "de_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "german_charfilter" ], "filter": [ "de_icu_normalizer", "german_stop", "german_normalization", "german_stemmer", "de_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "en_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "en_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "en_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "possessive_english", "icu_normalizer", "stop", "icu_folding", "remove_empty", "kstem", "custom_stem" ], "tokenizer": "standard", "type": "custom" }, "en_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "possessive_english", "icu_normalizer", "stop", "icu_folding", "remove_empty", "kstem", "custom_stem" ], "tokenizer": "standard", "type": "custom" }, "eo_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "eo_icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "eo_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "eo_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "eo_icu_folding", "remove_empty", "esperanto_stemmer" ], "tokenizer": "standard", "type": "custom" }, "eo_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "eo_icu_folding", "remove_empty", "esperanto_stemmer" ], "tokenizer": "standard", "type": "custom" }, "es_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "es_icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "es_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "es_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "spanish_stop", "spanish_stemmer", "es_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "es_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "spanish_stop", "spanish_stemmer", "es_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "he_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "he_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "he_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "niqqud", "hebrew_lemmatizer", "remove_duplicates", "icu_normalizer", "icu_folding", "remove_empty" ], "tokenizer": "hebrew", "type": "custom" }, "he_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "niqqud", "hebrew_lemmatizer", "remove_duplicates", "icu_normalizer", "icu_folding", "remove_empty" ], "tokenizer": "hebrew", "type": "custom" }, "km_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "km_icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "km_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "km_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "khmer_syll_reorder", "khmer_numbers" ], "filter": [ "icu_normalizer", "km_icu_folding", "remove_empty" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "km_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "khmer_syll_reorder", "khmer_numbers" ], "filter": [ "icu_normalizer", "km_icu_folding", "remove_empty" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "ko_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "ko_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "ko_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "arabic_extended_norm", "nori_charfilter", "nori_combo_filter" ], "filter": [ "nori_posfilter", "nori_readingform", "icu_normalizer", "icu_folding", "remove_empty" ], "tokenizer": "nori_tok", "type": "custom" }, "ko_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "arabic_extended_norm", "nori_charfilter", "nori_combo_filter" ], "filter": [ "nori_posfilter", "nori_readingform", "icu_normalizer", "icu_folding", "remove_empty" ], "tokenizer": "nori_tok", "type": "custom" }, "pl_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "pl_icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "pl_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "pl_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "polish_stop", "polish_stem", "stempel_pattern_filter", "pl_icu_folding", "remove_empty", "stempel_stop" ], "tokenizer": "standard", "type": "custom" }, "pl_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "polish_stop", "polish_stem", "stempel_pattern_filter", "pl_icu_folding", "remove_empty", "stempel_stop" ], "tokenizer": "standard", "type": "custom" }, "ru_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper", "russian_charfilter" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "ru_icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "ru_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper", "russian_charfilter" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "ru_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "russian_charfilter" ], "filter": [ "icu_normalizer", "russian_stop", "russian_stemmer", "ru_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "ru_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "russian_charfilter" ], "filter": [ "icu_normalizer", "russian_stop", "russian_stemmer", "ru_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "sk_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "sk_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "sk_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "slovak_stemmer", "icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "sk_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "slovak_stemmer", "icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "sr_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "sr_icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "sr_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "sr_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "sr_icu_folding", "remove_empty", "serbian_stemmer" ], "tokenizer": "standard", "type": "custom" }, "sr_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "icu_normalizer", "sr_icu_folding", "remove_empty", "serbian_stemmer" ], "tokenizer": "standard", "type": "custom" }, "tr_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "lowercase", "icu_normalizer", "preserve_original_recorder", "tr_icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "tr_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "lowercase", "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "tr_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "lowercase", "icu_normalizer", "better_apostrophe", "turkish_stop", "turkish_stemmer", "tr_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "tr_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm" ], "filter": [ "lowercase", "icu_normalizer", "better_apostrophe", "turkish_stop", "turkish_stemmer", "tr_icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "uk_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer", "preserve_original_recorder", "icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "uk_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "icu_normalizer" ], "tokenizer": "standard", "type": "custom" }, "uk_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "ukrainian_charfilter" ], "filter": [ "icu_normalizer", "ukrainian_stop", "ukrainian_stemmer", "icu_normalizer", "remove_duplicates", "icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "uk_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "word_break_helper", "arabic_extended_norm", "ukrainian_charfilter" ], "filter": [ "icu_normalizer", "ukrainian_stop", "ukrainian_stemmer", "icu_normalizer", "remove_duplicates", "icu_folding", "remove_empty" ], "tokenizer": "standard", "type": "custom" }, "zh_plain": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "smartcn_stop", "icu_normalizer", "preserve_original_recorder", "icu_folding", "preserve_original", "remove_empty" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "zh_plain_search": { "char_filter": [ "nnbsp_norm", "word_break_helper" ], "filter": [ "smartcn_stop", "icu_normalizer" ], "tokenizer": "icu_tokenizer", "type": "custom" }, "zh_text": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "dotted_I_fix", "arabic_extended_norm", "stconvertfix", "tsconvert" ], "filter": [ "smartcn_stop", "icu_normalizer" ], "tokenizer": "smartcn_tokenizer", "type": "custom" }, "zh_text_search": { "char_filter": [ "globo_norm", "regex_acronym_fixer", "regex_camelCase", "dotted_I_fix", "arabic_extended_norm", "stconvertfix", "tsconvert" ], "filter": [ "smartcn_stop", "icu_normalizer" ], "tokenizer": "smartcn_tokenizer", "type": "custom" } }, "char_filter": { "arabic_extended_norm": { "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" ], "type": "mapping" }, "dotted_I_fix": { "mappings": [ "\u0130=>I" ], "type": "mapping" }, "german_charfilter": { "mappings": [ "\u1e9e=>\u00df" ], "type": "mapping" }, "globo_norm": { "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" ], "type": "mapping" }, "khmer_numbers": { "mappings": [ "\\u17e0=>0", "\\u17e1=>1", "\\u17e2=>2", "\\u17e3=>3", "\\u17e4=>4", "\\u17e5=>5", "\\u17e6=>6", "\\u17e7=>7", "\\u17e8=>8", "\\u17e9=>9" ], "type": "mapping" }, "nnbsp_norm": { "mappings": [ "\\u202F=>\\u0020" ], "type": "mapping" }, "nori_charfilter": { "mappings": [ "\\u00B7=>\\u0020", "\\u318D=>\\u0020", "\\u00AD=>", "\\u200C=>" ], "type": "mapping" }, "nori_combo_filter": { "pattern": "[\\u0300-\\u0331]", "replacement": "", "type": "pattern_replace" }, "regex_acronym_fixer": { "pattern": "(?<=(?:^|\\P{L})\\p{L})[.\uff0e](\\p{L})(?=\\P{L}|$)", "replacement": "$1", "type": "pattern_replace" }, "regex_camelCase": { "pattern": "(\\p{Ll}[\\p{M}\\p{Cf}]*)([\\p{Lu}\\p{Lt}])", "replacement": "$1 $2", "type": "pattern_replace" }, "russian_charfilter": { "mappings": [ "\\u0301=>", "\\u0435\\u0308=>\\u0435", "\\u0415\\u0308=>\\u0415", "\\u0451=>\\u0435", "\\u0401=>\\u0415" ], "type": "mapping" }, "stconvertfix": { "mappings": [ "\\u606d\\u5f18=>\\u606d \\u5f18", "\\u5138=>\\u3469" ], "type": "mapping" }, "tsconvert": { "convert_type": "t2s", "delimiter": "#", "keep_both": false, "type": "stconvert" }, "ukrainian_charfilter": { "mappings": [ "\u2018=>'", "\u2019=>'", "`=>'", "\u00b4=>'", "\u02bc=>'", "\\u0301=>", "\\u00AD=>", "\u0491=>\u0433", "\u0490=>\u0413" ], "type": "mapping" }, "word_break_helper": { "mappings": [ "_=>\\u0020", ":=>\\u0020", ".=>\\u0020", "(=>\\u0020", ")=>\\u0020", "\uff0e=>\\u0020", "\uff3f=>\\u0020", "\uff1a=>\\u0020", "\u00b7=>\\u0020" ], "type": "mapping" } }, "filter": { "custom_stem": { "rules": "guidelines => guideline", "type": "stemmer_override" }, "de_icu_folding": { "type": "icu_folding", "unicodeSetFilter": "[^\u00c4\u00e4\u00d6\u00f6\u00dc\u00fc\u1e9e\u00df]" }, "de_icu_normalizer": { "name": "nfkc_cf", "type": "icu_normalizer", "unicodeSetFilter": "[^\u1e9e\u00df]" }, "eo_icu_folding": { "type": "icu_folding", "unicodeSetFilter": "[^\u0108\u0109\u011c\u011d\u0124\u0125\u0134\u0135\u015c\u015d\u016c\u016d]" }, "es_icu_folding": { "type": "icu_folding", "unicodeSetFilter": "[^\u00d1\u00f1]" }, "german_stemmer": { "language": "light_german", "type": "stemmer" }, "german_stop": { "stopwords": "_german_", "type": "stop" }, "icu_folding": { "type": "icu_folding" }, "icu_normalizer": { "name": "nfkc_cf", "type": "icu_normalizer" }, "km_icu_folding": { "type": "icu_folding", "unicodeSetFilter": "[^\u1780-\u17dd]" }, "lowercase": { "language": "turkish", "type": "lowercase" }, "nori_posfilter": { "stoptags": [ "E", "IC", "J", "MAG", "MAJ", "MM", "SP", "SSC", "SSO", "SC", "SE", "XPN", "XSA", "XSN", "XSV", "UNA", "NA", "VSV", "VCP", "VCN", "VX" ], "type": "nori_part_of_speech" }, "pl_icu_folding": { "type": "icu_folding", "unicodeSetFilter": "[^\u0104\u0105\u0106\u0107\u0118\u0119\u0141\u0142\u0143\u0144\u00d3\u00f3\u015a\u015b\u0179\u017a\u017b\u017c]" }, "polish_stop": { "stopwords": [ "vol", "o.o.", "o.o", "mgr", "godz", "z\u0142", "www", "pl", "ul", "tel", "hab", "prof", "in\u017c", "dr", "i", "u", "aby", "albo", "ale", "ani", "a\u017c", "bardzo", "bez", "bo", "bowiem", "by", "byli", "bym", "by\u0142", "by\u0142a", "by\u0142o", "by\u0142y", "by\u0107", "b\u0119dzie", "b\u0119d\u0105", "chce", "cho\u0107", "co", "coraz", "co\u015b", "czy", "czyli", "cz\u0119sto", "dla", "do", "gdy", "gdyby", "gdy\u017c", "gdzie", "go", "ich", "im", "inne", "i\u017c", "ja", "jak", "jakie", "jako", "je", "jednak", "jednym", "jedynie", "jego", "jej", "jest", "jeszcze", "je\u015bli", "je\u017celi", "ju\u017c", "j\u0105", "kiedy", "kilku", "kto", "kt\u00f3ra", "kt\u00f3re", "kt\u00f3rego", "kt\u00f3rej", "kt\u00f3ry", "kt\u00f3rych", "kt\u00f3rym", "kt\u00f3rzy", "lat", "lecz", "lub", "ma", "maj\u0105", "mamy", "mi", "mia\u0142", "mimo", "mnie", "mog\u0105", "mo\u017ce", "mo\u017cna", "mu", "musi", "na", "nad", "nam", "nas", "nawet", "nic", "nich", "nie", "niej", "nim", "ni\u017c", "no", "nowe", "np", "nr", "o", "od", "ok", "on", "one", "oraz", "pan", "po", "pod", "ponad", "poniewa\u017c", "poza", "przed", "przede", "przez", "przy", "raz", "razie", "roku", "r\u00f3wnie\u017c", "si\u0119", "sobie", "spos\u00f3b", "swoje", "s\u0105", "ta", "tak", "takich", "takie", "tak\u017ce", "tam", "te", "tego", "tej", "temu", "ten", "teraz", "te\u017c", "to", "trzeba", "tu", "tych", "tylko", "tym", "tys", "tzw", "t\u0119", "w", "we", "wie", "wi\u0119c", "wszystko", "w\u015br\u00f3d", "w\u0142a\u015bnie", "z", "za", "za\u015b", "ze", "\u017ce", "\u017ceby", "ii", "iii", "iv", "vi", "vii", "viii", "ix", "xi", "xii", "xiii", "xiv", "xv" ], "type": "stop" }, "possessive_english": { "language": "possessive_english", "type": "stemmer" }, "remove_empty": { "min": 1, "type": "length" }, "ru_icu_folding": { "type": "icu_folding", "unicodeSetFilter": "[^\u0419\u0439]" }, "russian_stemmer": { "language": "russian", "type": "stemmer" }, "russian_stop": { "stopwords": "_russian_", "type": "stop" }, "smartcn_stop": { "stopwords": [ "," ], "type": "stop" }, "spanish_stemmer": { "language": "light_spanish", "type": "stemmer" }, "spanish_stop": { "stopwords": "_spanish_", "type": "stop" }, "sr_icu_folding": { "type": "icu_folding", "unicodeSetFilter": "[^\u0110\u0111\u017d\u017e\u0106\u0107\u0160\u0161\u010c\u010d]" }, "stempel_pattern_filter": { "pattern": "^([a-z\u0105\u0142]?[a-z\u0107\u0144]|..\u0107|\\d.*\u0107)$", "replacement": "", "type": "pattern_replace" }, "stempel_stop": { "stopwords": [ "owa\u0107", "iwa\u0107", "obi\u0107", "sn\u0105\u0107", "ywa\u0107", "ium", "my", "um" ], "type": "stop" }, "tr_icu_folding": { "type": "icu_folding", "unicodeSetFilter": "[^\u00c7\u00e7\u011e\u011fI\u0131\u0130i\u00d6\u00f6\u015e\u015f\u00dc\u00fc]" }, "turkish_stemmer": { "language": "turkish", "type": "stemmer" }, "turkish_stop": { "stopwords": "_turkish_", "type": "stop" } }, "tokenizer": { "nori_tok": { "decompound_mode": "mixed", "type": "nori_tokenizer" } } }