org.languagetool.resource.br.disambiguation.xml Maven / Gradle / Ivy
The newest version!
<?xml version="1.0" encoding="utf-8"?> <!-- Breton Disambiguation Rules for LanguageTool Copyright (C) 2011-2014 Dominique Pellé <[email protected]> --> <rules lang="br" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:noNamespaceSchemaLocation="../../../../../../../../../languagetool-core/src/main/resources/org/languagetool/resource/disambiguation.xsd"> <rulegroup id="EN_UR" name="en ur + V"> <rule> <pattern> <token>en</token> <marker> <token>ur</token> </marker> <token postag="V inf.*" postag_regexp="yes"/> </pattern> <disambig postag="X EN_UR"/> </rule> <rule> <pattern> <token regexp="yes">[’'‘]</token> <token>n</token> <marker> <token>ur</token> </marker> <token postag="V inf.*" postag_regexp="yes"/> </pattern> <disambig postag="X EN_UR"/> </rule> <rule> <pattern> <token>en</token> <marker> <token>ur</token> </marker> <token postag="UNKNOWN" regexp="yes">.*[ai]ñ</token> </pattern> <disambig postag="X EN_UR"/> </rule> </rulegroup> <rulegroup id="XXI" name="Roman Number"> <rule> <pattern case_sensitive="yes"> <token regexp="yes">[MCDXLVI]+</token> </pattern> <disambig postag="K e sp"/> </rule> <rule> <pattern case_sensitive="yes"> <token regexp="yes">[MCDXLVI]+-?vet|I-?añ</token> </pattern> <disambig postag="K e sp o"/> </rule> </rulegroup> <rulegroup id="PREP_A" name="preposition a"> <rule> <pattern> <token regexp="yes">a-walc’h|bep|(e-)?leizh|kalz(ig)?|kement|lies|meur|muioc’h|nebeut(oc’h)?|nemeur|niver|pep|re|trawalc’h</token> <marker> <token>a</token> </marker> <token postag="N.*" postag_regexp="yes"/> </pattern> <disambig postag="P"/> </rule> <rule> <pattern> <token postag="K . p" postag_regexp="yes"/> <marker> <token>a</token> </marker> </pattern> <disambig postag="P"/> </rule> <rule> <pattern> <token>pegement</token> <marker><token>a</token></marker> </pattern> <disambig action="filter" postag="[LP].*"/> </rule> <rule> <pattern> <token regexp="yes">u[nr]</token> <token postag="J.*" postag_regexp="yes"/> <marker><token>a</token></marker> <token postag="N.*" postag_regexp="yes"/> </pattern> <disambig action="filter" postag="[LP].*"/> </rule> </rulegroup> <rulegroup id="O" name="o"> <rule id="O_R" name="O pronoun"> <pattern> <marker><token>o</token></marker> <token postag="N .*" postag_regexp="yes"><exception postag="V inf.*" postag_regexp="yes"/></token> </pattern> <disambig postag="R e p 3 obj"/> </rule> <rule> <pattern> <marker><token>o</token></marker> <token postag="V inf.*" postag_regexp="yes"><exception postag="N.*" postag_regexp="yes"/></token> </pattern> <disambig postag="L o"/> </rule> <rule> <pattern> <token postag="SENT_START"/> <token postag="[VNJZ].*|R suj.*" postag_regexp="yes"/> <marker> <token>a</token> </marker> <token postag="V.*" postag_regexp="yes"/> </pattern> <disambig postag="L a"/> </rule> <rule> <pattern> <token postag="[NJ].*" postag_regexp="yes"/> <token>hag</token> <marker><token>a</token></marker> <token postag="V.*" postag_regexp="yes"/> </pattern> <disambig postag="L a"/> </rule> <rule> <pattern> <token>peogwir</token> <marker> <token>e</token> </marker> </pattern> <disambig postag="L e"/> </rule> <rule> <pattern> <token>peogwir</token> <marker> <token>e</token> </marker> </pattern> <disambig postag="L e"/> </rule> <rule> <pattern> <marker> <token>e</token> </marker> <token postag="V.*" postag_regexp="yes"><exception postag="[ANZ].*|V inf.*" postag_regexp="yes"/></token> </pattern> <disambig postag="L e"/> </rule> <rule> <pattern> <marker> <token>e</token> </marker> <token>vez</token> </pattern> <disambig postag="L e"/> </rule> <rule> <pattern> <token postag="SENT_START"/> <token postag="D.*" postag_regexp="yes"/> <token postag="N.*" postag_regexp="yes"/> <marker> <token>a</token> </marker> <token postag="V.*" postag_regexp="yes"/> </pattern> <disambig postag="L a"/> </rule> <rule> <pattern> <token postag="V inf"/> <token>a</token> <token inflected="yes">ober</token> <marker><token>o</token></marker> <token postag="V inf.*" postag_regexp="yes"/> </pattern> <disambig postag="R e p 3 obj"/> </rule> </rulegroup> <rulegroup id="NA" name="na"> <rule> <pattern> <token postag="SENT_START"/> <marker><token>na</token></marker> <token postag="V impe .*" postag_regexp="yes"/> <token>ket</token> </pattern> <disambig postag="L n"/> </rule> <rule> <pattern> <token regexp="yes">aon|ha|betek-gouzout|evit|gant</token> <marker><token>na</token></marker> <token postag="V .*" postag_regexp="yes"/> <token>ket</token> </pattern> <disambig postag="L n"/> </rule> <rule> <pattern> <token regexp="yes">diwall(it)?|ken|hep|eget|evezh</token> <marker><token>na</token></marker> <token postag="V .*" postag_regexp="yes"/> </pattern> <disambig postag="L n"/> </rule> <rule> <pattern> <token>evit</token> <token>mirout</token> <marker><token>na</token></marker> <token postag="V .*" postag_regexp="yes"/> </pattern> <disambig postag="L n"/> </rule> <rule> <pattern> <token postag="SENT_START"/> <marker><token>na</token></marker> <token postag="J exc.*" postag_regexp="yes"/> </pattern> <disambig postag="A pre"/> </rule> <rule> <pattern> <token postag="SENT_START"/> <marker><token>na</token></marker> <token postag="[JE].*" postag_regexp="yes"/> <token regexp="yes">eo?</token> </pattern> <disambig postag="A pre"/> </rule> <rule> <pattern> <marker><token>na</token></marker> <token postag="[NJ].*|V inf.*" postag_regexp="yes"/> <token regexp="yes">nag?</token> </pattern> <disambig postag="C coor"/> </rule> <rule> <pattern> <marker><token>na</token></marker> <token postag="[NJ].*|V inf.*" postag_regexp="yes"/> <token>,</token> <token regexp="yes">nag?</token> </pattern> <disambig postag="C coor"/> </rule> <rule> <pattern> <token regexp="yes">na|hep</token> <token postag="[NJ].*|V inf.*" postag_regexp="yes"/> <marker><token>na</token></marker> </pattern> <disambig postag="C coor"/> </rule> <rule> <pattern> <token regexp="yes">na|hep</token> <token postag="[NJ].*|V inf.*" postag_regexp="yes"/> <token>,</token> <marker><token>na</token></marker> </pattern> <disambig postag="C coor"/> </rule> <rule> <pattern> <token regexp="yes">n[e’]</token> <token postag="V.*" postag_regexp="yes"/> <marker><token>na</token></marker> </pattern> <disambig postag="C coor"/> </rule> <rule> <pattern> <token regexp="yes">n[e’]</token> <token postag="V.*" postag_regexp="yes"/> <token>,</token> <marker><token>na</token></marker> </pattern> <disambig postag="C coor"/> </rule> </rulegroup> <rule id="D_O" name="d’o"> <pattern> <token>d’</token> <marker><token>o</token></marker> </pattern> <disambig postag="R e p 3 obj"/> </rule> <rule id="D_E" name="d’e"> <pattern> <token>d’</token> <marker><token>e</token></marker> </pattern> <disambig postag="R m s 1 obj"/> </rule> <rulegroup id="EZ_AN" name="ez an"> <rule> <pattern> <marker> <token regexp="yes">e?z|ned</token> </marker> <token regexp="yes">an|in|aen|afen|ajen|ez|i|aes|afes|ajes|a|aio|ae|afe|aje|eomp|aimp|aemp|afemp|ajemp|it|eot|eoc’h|aec’h|afec’h|ajec’h|eont|aint|aent|afent|ajent|eer|eor|aed|afed|ajed</token> </pattern> <disambig postag="L e"/> </rule> <rule> <pattern> <token>ez</token> <marker> <token>an</token> </marker> </pattern> <disambig action="remove"><wd lemma="an" pos="D e sp"/></disambig> </rule> </rulegroup> <rulegroup id="EN_EM" name="en em"> <rule> <pattern> <marker><token regexp="yes">e?n</token></marker> <token>em</token> </pattern> <disambig postag="X EN_EM"/> </rule> <rule> <pattern> <token regexp="yes">e?n</token> <marker><token>em</token></marker> </pattern> <disambig postag="X EN_EM"/> </rule> </rulegroup> <rulegroup id="EZ_A" name="ez a"> <rule> <pattern> <token regexp="yes">e?z</token> <marker> <token>a</token> </marker> </pattern> <disambig><wd lemma="mont" pos="V pres 3 s"/></disambig> </rule> </rulegroup> <rulegroup id="MA" name="ma"> <rule> <pattern> <marker> <token>ma</token> </marker> <token regexp="yes">digarez(it)?</token> </pattern> <disambig postag="R e s 1 obj"/> </rule> <rule> <pattern> <marker> <token>ma</token> </marker> <token postag="N.*" postag_regexp="yes"><exception postag="V.*" postag_regexp="yes"/></token> </pattern> <disambig postag="D e sp"/> </rule> <rule> <pattern> <marker> <token>ma</token> </marker> <token regexp="yes">.*-me</token> </pattern> <disambig postag="D e sp"/> </rule> <rule> <pattern> <marker> <token>ma</token> </marker> <token postag="V.*" postag_regexp="yes"><exception postag="N.*" postag_regexp="yes"/><exception regexp="yes">digarez(it)?</exception></token> </pattern> <disambig postag="C sub"/> </rule> <rule> <pattern> <marker> <token>ma</token> </marker> <token postag="V inf.*" postag_regexp="yes"/> <token>a</token> <token inflected="yes">ober</token> </pattern> <disambig postag="R e s 1 obj"/> </rule> </rulegroup> <rulegroup id="KAOUT" name="kaout"> <rule> <pattern> <marker> <token regexp="yes">[mz]|[ea]([mnz]|c’h)|he|ho[nr]?|hoc’h|o</token> </marker> <token regexp="yes">[pd]?eus|[bpd](e[fz][eo]|ije|ez|o[ae]?)|deveze?|devo[ae]</token> </pattern> <disambig postag="X KAOUT1"/> </rule> <rule> <pattern> <token regexp="yes">[mz]|[ea]([mnz]|c’h)|he|ho[nr]?|hoc’h|o</token> <marker> <token regexp="yes">[pd]?eus|[bpd](e[fz][eo]|ije|ez|o[ae]?)|deveze?|devo[ae]</token> </marker> </pattern> <disambig postag="X KAOUT2"/> </rule> </rulegroup> <rulegroup id="E_R" name="e pronom"> <rule> <pattern> <marker><token>e</token></marker> <token postag="V inf.*" postag_regexp="yes"/> <token>a</token> <token inflected="yes">ober</token> </pattern> <disambig action="remove"><wd lemma="e" pos="P"/></disambig> </rule> <rule> <pattern> <marker><token>E</token></marker> <token postag="V inf.*" postag_regexp="yes"/> <token>a</token> <token inflected="yes">ober</token> </pattern> <disambig action="remove"><wd lemma="e" pos="L e"/></disambig> </rule> <rule> <pattern> <token regexp="yes">dirak|en|e-pad|eus|evit|(di)?(gant|ouzh|war)<exception postag="X KAOUT2"/></token> <marker><token>e</token></marker> </pattern> <disambig action="remove"><wd lemma="e" pos="P"/></disambig> </rule> <rule> <pattern> <token regexp="yes">dirak|en|e-pad|eus|evit|(di)?(gant|ouzh|war)<exception postag="X KAOUT2"/></token> <marker><token>e</token></marker> </pattern> <disambig action="remove"><wd lemma="e" pos="L e"/></disambig> </rule> <rule> <pattern> <token postag="SENT_START|C coor" postag_regexp="yes"/> <marker><token>e</token></marker> <token postag="N.*" postag_regexp="yes"/> <token>zo</token> <token postag="J"/> </pattern> <disambig action="remove"><wd lemma="e" pos="L e"/></disambig> </rule> <rule> <pattern> <marker><token>e</token></marker> <token postag="N.*" postag_regexp="yes"/> <token regexp="yes">[a’]</token> <token>zo</token> </pattern> <disambig action="remove"><wd lemma="e" pos="P"/></disambig> </rule> <rule> <pattern> <marker><token>e</token></marker> <token postag="N.*" postag_regexp="yes"/> <token regexp="yes">[a’]</token> <token>zo</token> </pattern> <disambig action="remove"><wd lemma="e" pos="L e"/></disambig> </rule> </rulegroup> <rulegroup id="E_P" name="e preposition"> <rule> <pattern> <token inflected="yes" regexp="yes">trein|lakaat|skrivañ</token> <marker> <token postag="P">e</token> </marker> <token postag_regexp="yes" postag="N.*" regexp="yes">.*eg</token> </pattern> <disambig postag="P"/> </rule> </rulegroup> <rulegroup id="NE_V" name="ne + verb"> <rule> <pattern> <token regexp="yes">ned?|n’</token> <marker> <token postag="V.*" postag_regexp="yes"/> </marker> </pattern> <disambig><match no="1" postag_regexp="yes" postag="V.*"/></disambig> </rule> </rulegroup> <rulegroup id="UR_N" name="ur + anv"> <rule> <pattern> <token>ur<exception scope="previous">en</exception></token> <marker> <token postag="V.*" postag_regexp="yes"/> </marker> </pattern> <disambig action="filter" postag="N.*"/> </rule> <rule> <pattern> <token regexp="yes">u[ln]|a[nlr]</token> <marker> <token postag="V.*" postag_regexp="yes"/> </marker> </pattern> <disambig action="filter" postag="N.*"/> <example type="untouched">An eil.</example> </rule> </rulegroup> <rule id="PE_INT" name="pe interrogatif"> <pattern> <token postag="SENT_START"/> <marker> <token>pe</token> </marker> <token postag="N.*" postag_regexp="yes"><exception>marteze</exception></token> </pattern> <disambig postag="J itg"/> </rule> <rule id="RA_L" name="particule ra"> <pattern> <token postag="SENT_START"/> <marker> <token>ra</token> </marker> <token postag="V (futu|conf) .*" postag_regexp="yes"/> </pattern> <disambig postag="L r"/> </rule> <rule id="RA_V" name="ra verbe"> <pattern> <token postag="V inf"/> <token>a</token> <marker><token>ra</token></marker> </pattern> <disambig postag="V pres 3 s"/> </rule> <rule id="KOZH" name="kozh"> <pattern> <token regexp="yes">([kg]|c’h)ozh</token> </pattern> <disambig action="filter" postag="J.*"/> </rule> <rule id="PAOT_MAT" name="paot mat"> <pattern> <marker> <token>paot</token> </marker> <token>mat</token> </pattern> <disambig action="filter" postag="J"/> </rule> <rule id="CHENTAN" name="c’hentañ"> <pattern> <token>ken</token> <token>ar</token> <marker> <token>c’hentañ</token> </marker> </pattern> <disambig action="filter" postag="J.*"/> </rule> <rule id="A_UNAN_DA_UNAN" name="a unan da unan"> <pattern> <token>a</token> <token>unan</token> <token>da</token> <marker> <token>unan</token> </marker> </pattern> <disambig postag="X A_UNAN_DA_UNAN"/> </rule> <rulegroup id="GANT" name="gant"> <rule> <pattern> <token>da</token> <marker><token>gant</token></marker> </pattern> <disambig action="remove"><wd lemma="gant" pos="P"/></disambig> </rule> <rule> <pattern> <token postag="SENT_START"/> <marker><token>gant</token></marker> </pattern> <disambig postag="P"/> </rule> <rule id="AZEN_GORNEK" name="azen gornek"> <pattern> <marker><token>azen</token></marker> <token>gornek</token> </pattern> <disambig postag="N f s"/> </rule> </rulegroup> <rule id="REZ" name="rez"> <pattern> <token postag="V inf"/> <token>a</token> <marker><token>rez</token></marker> </pattern> <disambig postag="V pres 2 s"/> </rule> <rulegroup id="FRANCE" name="France"> <rule> <pattern case_sensitive="yes"> <marker><token>France</token></marker> <token regexp="yes">[23]|Bleu|Culture|Info|Inter|Musique|Télécom|Télévisions</token> </pattern> <disambig action="immunize"/> </rule> <rule> <pattern case_sensitive="yes"> <token regexp="yes">Air|Anatole|Radio|de</token> <marker><token>France</token></marker> </pattern> <disambig action="immunize"/> </rule> </rulegroup> <rule id="OBER_VAN" name="ober van"> <pattern> <token inflected="yes">ober</token> <marker><token>van</token></marker> </pattern> <disambig action="immunize"/> </rule> <rule id="KENTAN" name="kentañ"> <pattern> <token>skol</token> <token>kentañ</token> <token>derez</token> </pattern> <disambig action="immunize"/> </rule> <rulegroup id="GERIENNOU_BENNAK" name="ur geriennoù bennak"> <rule> <pattern> <token>ur</token> <token>geriennoù</token> <token>bennak</token> </pattern> <disambig action="immunize"/> </rule> <rule> <pattern> <token>un</token> <token regexp="yes">deizioù|hiniennoù</token> <token>bennak</token> </pattern> <disambig action="immunize"/> </rule> </rulegroup> <rulegroup id="DA_LAVARET_EO" name="Da lavaret eo"> <rule> <pattern> <token>da</token> <token regexp="yes">lavaret|lavarout|lâret</token> <marker><token>eo</token></marker> </pattern> <disambig action="add"><wd lemma="bezañ" pos="X DA_LAVARET_EO"/></disambig> </rule> <rule> <pattern> <token>d</token> <token>.</token> <token>l</token> <token>.</token> <marker><token>e</token></marker> </pattern> <disambig><wd lemma="bezañ" pos="X DA_LAVARET_EO"/></disambig> </rule> </rulegroup> </rules>