From 9541c8e5dda00dcd7a7fb98a8e90d9b12aa45a54 Mon Sep 17 00:00:00 2001 From: Al Date: Thu, 9 Jul 2015 18:47:48 -0400 Subject: [PATCH] [dictionaries] no longer assuming that we'll be stripping internal periods in non-acronyms e.g. v.le for viale in Italian --- data/dictionaries/ca/no_number.txt | 2 +- data/dictionaries/ca/place_names.txt | 4 +- data/dictionaries/ca/unit_types.txt | 2 +- data/dictionaries/de/stopwords.txt | 8 +- data/dictionaries/es/company_types.txt | 58 +++--- data/dictionaries/es/directionals.txt | 8 +- data/dictionaries/es/given_names.txt | 2 +- data/dictionaries/es/no_number.txt | 2 +- data/dictionaries/es/personal_titles.txt | 54 +++--- data/dictionaries/es/place_names.txt | 4 +- data/dictionaries/es/qualifiers.txt | 4 +- data/dictionaries/es/street_types.txt | 8 +- data/dictionaries/es/unit_types.txt | 6 +- data/dictionaries/fr/company_types.txt | 52 +++--- data/dictionaries/fr/directionals.txt | 6 +- data/dictionaries/fr/personal_titles.txt | 14 +- data/dictionaries/fr/place_names.txt | 15 +- data/dictionaries/fr/post_office.txt | 2 +- data/dictionaries/fr/street_types.txt | 6 +- data/dictionaries/it/company_types.txt | 26 +-- data/dictionaries/it/directionals.txt | 4 +- data/dictionaries/it/personal_titles.txt | 26 +-- data/dictionaries/it/place_names.txt | 174 +++++++++--------- data/dictionaries/it/post_office.txt | 2 +- data/dictionaries/it/qualifiers.txt | 18 +- data/dictionaries/it/stopwords.txt | 8 +- data/dictionaries/it/street_types.txt | 96 +++++----- data/dictionaries/it/synonyms.txt | 194 ++++++++++----------- data/dictionaries/it/unit_types.txt | 22 +-- data/dictionaries/nl/directionals.txt | 8 +- data/dictionaries/nl/qualifiers.txt | 8 +- data/dictionaries/pt/company_types.txt | 52 +++--- data/dictionaries/pt/directionals.txt | 8 +- data/dictionaries/pt/no_number.txt | 2 +- data/dictionaries/pt/personal_suffixes.txt | 2 +- data/dictionaries/pt/personal_titles.txt | 80 ++++----- data/dictionaries/pt/place_names.txt | 26 +-- data/dictionaries/pt/qualifiers.txt | 2 +- data/dictionaries/pt/stopwords.txt | 2 +- data/dictionaries/pt/synonyms.txt | 18 +- data/dictionaries/pt/toponyms.txt | 2 +- data/dictionaries/pt/unit_types.txt | 2 +- 42 files changed, 522 insertions(+), 517 deletions(-) diff --git a/data/dictionaries/ca/no_number.txt b/data/dictionaries/ca/no_number.txt index 08e62eec..470b6059 100644 --- a/data/dictionaries/ca/no_number.txt +++ b/data/dictionaries/ca/no_number.txt @@ -1 +1 @@ -sense numero|s / n|s n|sn \ No newline at end of file +sense numero|s / n|s.n.|s.n|s n|sn \ No newline at end of file diff --git a/data/dictionaries/ca/place_names.txt b/data/dictionaries/ca/place_names.txt index 76b122b8..04737d2a 100644 --- a/data/dictionaries/ca/place_names.txt +++ b/data/dictionaries/ca/place_names.txt @@ -1,7 +1,7 @@ biblioteca|bibl cementiri|cem -collegi|col legi|col l|coll -collegi public|col legi public|cp +collegi|col·legi|col·l|coll +collegi public|col·legi public|cp comissaria|com convent|convt correus|corr diff --git a/data/dictionaries/ca/unit_types.txt b/data/dictionaries/ca/unit_types.txt index 7fa76fbf..ce21980b 100644 --- a/data/dictionaries/ca/unit_types.txt +++ b/data/dictionaries/ca/unit_types.txt @@ -7,6 +7,6 @@ entrada|entr escala|esc habitacio|hab numero|num|n|# -parcella|parcel la|parc +parcella|parcel·la|parc porta|pta sobreatic|s / at|s at \ No newline at end of file diff --git a/data/dictionaries/de/stopwords.txt b/data/dictionaries/de/stopwords.txt index 5326dbd9..a6fe0488 100644 --- a/data/dictionaries/de/stopwords.txt +++ b/data/dictionaries/de/stopwords.txt @@ -1,8 +1,8 @@ am an -an der|a.d.|a d +an der|a.d.|a.d|a d auf -auf der|a.d.|a d +auf der|a.d.|a.d|a d bei|b beim bis @@ -18,7 +18,7 @@ fur gegenuber im|i in|i -in der|i d +in der|i.d.|i.d|i d mit nach nachst @@ -30,7 +30,7 @@ und|& unter vor|v von|v -von der|v d +von der|v.d.|v.d|v d zu zu der zur diff --git a/data/dictionaries/es/company_types.txt b/data/dictionaries/es/company_types.txt index 99f1dc53..fe44aa86 100644 --- a/data/dictionaries/es/company_types.txt +++ b/data/dictionaries/es/company_types.txt @@ -1,48 +1,48 @@ asociacion en participacion|a en p asociacion civil|ac|a c -banco|bco +banco|bco|b.co club atletico -club de futbol|cf|c f|futbol club|fc|f c -compania|cia|c ia|ca|c a|comp +club de futbol|cf|c f|c.f.|c.f|futbol club|fc|f.c.|f.c|f c +compania|cia|c.ia|ca|c.a|comp compania anonima|ca|c a compania por acciones|c por a|cxa|c x a empresa|empr|emp empresa individual de responsabilidad limitada|eirl|e i r l -empresa publica|ep|e p -empresa unipersonal|e u -sociedad|sdad|soc -sociedad anonima|sa|s a|sdad anon -sociedad anonima abierta|saa|s a a -sociedad anonima bursatil|sab|s a b -sociedad anonima deportiva|sad|s a d +empresa publica|ep|e.p.|e.p|e p +empresa unipersonal|e.u.|e.u|e u +sociedad|sdad|s.dad|soc +sociedad anonima|sa|s.a.|s.a|s a|sdad anon +sociedad anonima abierta|saa|s.a.a.|s.a.a|s a a +sociedad anonima bursatil|sab|s.a.b.|s.a.b|s a b +sociedad anonima deportiva|sad|s.a.d.|s.a.d|s a d sociedad anonima de capital variable|sa de cv|s a de c v -sociedad anonima cooperativa catalana limitada|sccl|s c c l -sociedad anonima espanola|sae -sociedad anonima financiera de inversion|safi|s a f i -sociedad anonima inscrita de capital abierto|saica|s a i c a -sociedad anonima laboral|sal -sociedad anonima promotora de inversion|sapi|s a p i -sociedad anonima simplificada|sas|s a s -sociedad civil|sc|s c +sociedad anonima cooperativa catalana limitada|sccl|s.c.c.l.|s.c.c.l|s c c l +sociedad anonima espanola|sae|s.a.e.|s.a.e|s a e +sociedad anonima financiera de inversion|safi|s.a.f.i.|s.a.f.i|s a f i +sociedad anonima inscrita de capital abierto|saica|s.a.i.c.a.|s.a.i.c.a|s a i c a +sociedad anonima laboral|sal|s.a.l.|s.a.l|s a l +sociedad anonima promotora de inversion|sapi|s.a.p.i.|s.a.p.i|s a p i +sociedad anonima simplificada|sas|s.a.s.|s.a.s|s a s +sociedad civil|sc|s.c.|s.c|s c sociedad colectiva|snc|s n c|soc cal|y cia sc|y cia s c|& cia sc|& cia s c sociedad comanditaria|scra|s cra sociedad cooperativa|scoop|s coop -sociedad cooperativa limitada|scl|s c l -sociedad de capital e industria|scel|s c e l -sociedad de garantia reciproca|sgr|s g r -sociedad de responsabilidad limitada|srl|s r l|s de rl|s de r l -sociedad del estado|se|s e +sociedad cooperativa limitada|scl|s.c.l.|s.c.l|s c l +sociedad de capital e industria|scel|s.c.e.l.|s.c.e.l|s c e l +sociedad de garantia reciproca|sgr|s.g.r.|s.g.r|s g r +sociedad de responsabilidad limitada|srl|s.r.l.|s.r.l|s r l|s de rl|s de r l +sociedad del estado|se|s.e.|s.e|s e sociedad en comandita|s en c|sc|y cia s en c|& cia s en c sociedad en comandita por acciones|s en c por a|sca|s c a|cia sca|cia s c a sociedad en nombre colectivo|cia|s en nc|s en n c sociedad en nombre colectivo de responsabilidad limitada|cia ltda|ltda sociedad gestora de instituciones de inversion colectiva|sgiic|s g i i c -sociedad limitada|sl|s l|sdad ltda -sociedad limitada laboral|sll|s l l -sociedad limitada nueva empresa|slne|s l n e -sociedad limitada unipersonal|slu|s l u -sociedad por acciones|spa|s p a -union deportiva|ud +sociedad limitada|sl|s l|s.l.|s.l|sdad ltda +sociedad limitada laboral|sll|s.l.l.|s.l.l|s l l +sociedad limitada nueva empresa|slne|s.l.n.e.|s.l.n.e|s l n e +sociedad limitada unipersonal|slu|s.l.u.|s.l.u|s l u +sociedad por acciones|spa|s.p.a.|s.p.a|s p a +union deportiva|ud|u.d.|u.d|u d union y compania|& compania|y cia|& cia y sucesores|& sucesores \ No newline at end of file diff --git a/data/dictionaries/es/directionals.txt b/data/dictionaries/es/directionals.txt index 092aab87..ca5d4270 100644 --- a/data/dictionaries/es/directionals.txt +++ b/data/dictionaries/es/directionals.txt @@ -1,10 +1,10 @@ este|e norte|n -noreste|nordeste|ne|n e -noroeste|nw|n w +noreste|nordeste|ne|n.e.|n.e|n e +noroeste|nw|n.w.|n.w|n w oeste|w oriente|ote poniente|pte sur|s -sureste|se|s e -suroeste|sw|s w \ No newline at end of file +sureste|se|s.e.|s.e|s e +suroeste|sw|s.w.|s.w|s w \ No newline at end of file diff --git a/data/dictionaries/es/given_names.txt b/data/dictionaries/es/given_names.txt index d3f2b568..34981095 100644 --- a/data/dictionaries/es/given_names.txt +++ b/data/dictionaries/es/given_names.txt @@ -1,3 +1,3 @@ federico|fco francisco|fco|franc -maria|ma|m a \ No newline at end of file +maria|ma|m.a \ No newline at end of file diff --git a/data/dictionaries/es/no_number.txt b/data/dictionaries/es/no_number.txt index bff4b0ac..3fafc658 100644 --- a/data/dictionaries/es/no_number.txt +++ b/data/dictionaries/es/no_number.txt @@ -1 +1 @@ -sin numero|s / n|s n|sn \ No newline at end of file +sin numero|s / n|s.n.|s.n|s n|sn \ No newline at end of file diff --git a/data/dictionaries/es/personal_titles.txt b/data/dictionaries/es/personal_titles.txt index 92b6c61e..7a70046b 100644 --- a/data/dictionaries/es/personal_titles.txt +++ b/data/dictionaries/es/personal_titles.txt @@ -1,8 +1,8 @@ abad|abd -abogada|abga|abgda|abg da -abogado|abg|abgdo|abg do -administrador|admor|adm or|admr -administradora|admora|adm ora|admra +abogada|abga|abgda|abg.da +abogado|abg|abgdo|abg.do +administrador|admor|adm.or|admr +administradora|admora|adm.ora|admra agrimensor|agrim alcalde|alcde alferez|alf|alfz @@ -17,9 +17,9 @@ baronesa brigadier|brg|bg|brig brigadier general|brg general|brg gral|brg genl|brg gnal|brg gn|bg general|bg gral|bg genl|bg gnal|bg gn|brig general|brig gral|brig genl|brig gnal|brig gn caballero|cab -cabo primero|cp|c p -cabo segundo|cs|c s -cabo tercero|ct|c t +cabo primero|cp|c.p.|c.p|c p +cabo segundo|cs|c.s.|c.s|c s +cabo tercero|ct|c.t.|c.t|c t cacique|cque capitan|cap|capt|ctn|cptn cardenal|card|cnal @@ -36,11 +36,11 @@ diacono|diac diacona|diaca diputado|dip director|dir -directora|dira|dir a +directora|dira|dir.a doctor|dr doctora|dra don|d -dona|da|dna|d a +dona|da|dna|d.a dragoneante|dg duque|dq duquesa|dqsa|dqa @@ -48,7 +48,7 @@ embajador|emb enfermera|enf|enfa escritor escritora -excelencia|exca|exc a +excelencia|exca|exc.a excelentisima|excma excelentisimo|excmo fray|fr @@ -68,10 +68,10 @@ inspector|insp intendente|int jefe|jf juez|jz -licenciada|lda|lica|lic a -licenciado|ldo|lic|lico|lic o +licenciada|lda|lica|lic.a +licenciado|ldo|lic|lico|lic.o locutor|loc -madre|me|m e +madre|me|m.e maestro|mstro|mtro majestad mariscal|mscal @@ -79,16 +79,16 @@ marques|mq|marqs marquesa|mqa|marqsa mayor|my|may mayor de brigada|may brig|my brig -mayor general|mg|m g|my gen|my gral +mayor general|mg|m.g.|m.g|m g|my gen|my gral ministerio|min ministro|mtro monsenor|mons|msnr notario|not nuestra senora|ns|n s|ntra sra obispo|ob -padre|pe|p e|pdre|p dre +padre|pe|p.e|pdre|p.dre piloto fluvial|plto flvial -presidenta|pdta|presa|pres a|presida|presid a|pta +presidenta|pdta|presa|pres.a|presida|presid.a|pta presidente|pdte|pres|presid|pte principe principal|pral @@ -108,19 +108,19 @@ sant|st santo|sto sargento|sarg|sgto|sargto sargento ayudante|sarg ay|sgto ay|sargto ay -sargento mayor|sm|s m|sarg my|sgto my|sargto my -sargento primero|sp|s p -sargento segundo|sg|s g -sargento viceprimero|sargento vice primero|sv|s v -secretario|secreto|secto -secretaria|secreta|secta +sargento mayor|sm|s.m.|s.m|s m|sarg my|sgto my|sargto my +sargento primero|sp|s.p.|s.p|s p +sargento segundo|sg|s.g.|s.g|s g +sargento viceprimero|sargento vice primero|sv|s.v.|s.v|s v +secretario|secreto|secto|sect.o +secretaria|secreta|secta|sect.a senador|sen senor|sr -senora|sra -senores|srs|sres -senorita|srta -serenisima|serma|ser ma -serenisimo|sermo|ser mo +senora|sra|s.ra +senores|srs|sres|s.res +senorita|srta|s.rta +serenisima|serma|ser.ma +serenisimo|sermo|ser.mo soldado|sold subteniente|subte|stn teniente|tn|tte diff --git a/data/dictionaries/es/place_names.txt b/data/dictionaries/es/place_names.txt index c783c1ec..df0378dc 100644 --- a/data/dictionaries/es/place_names.txt +++ b/data/dictionaries/es/place_names.txt @@ -5,7 +5,7 @@ arboleda|arb arrabal|arral arroyo|arry asuntos exteriores|aa ee -atletico|atl o|atlo +atletico|atlo ayuntamiento balneario|balnr banco @@ -41,7 +41,7 @@ estacion|estcn estacionamiento|estto fabrica|fca|fbrca farmacia -ferrocarril|fc|f c +ferrocarril|fc|f.c.|f.c|f c ferrocarriles|ff cc|ffcc granja|granj hacienda|hda diff --git a/data/dictionaries/es/qualifiers.txt b/data/dictionaries/es/qualifiers.txt index 7b8a4681..5f7a4c8d 100644 --- a/data/dictionaries/es/qualifiers.txt +++ b/data/dictionaries/es/qualifiers.txt @@ -11,8 +11,8 @@ estado|edo isla|isl|is islas|isls|iss poligono|polig -poligono residencial|polig res|pgres|pg res|pr|p r -poligono industrial|polig ind|pgind|pg ind|pi|p i +poligono residencial|polig res|pgres|pg res|pr|p.r.|p.r|p r +poligono industrial|polig ind|pgind|pg ind|pi|p.i.|p.i|p i seccion|secc|sccn|scn|sec sector|sect urbanizacion|urb diff --git a/data/dictionaries/es/street_types.txt b/data/dictionaries/es/street_types.txt index bfb7054c..a0e11337 100644 --- a/data/dictionaries/es/street_types.txt +++ b/data/dictionaries/es/street_types.txt @@ -25,9 +25,9 @@ callejuela|cjla callizo|cllzo calzada|czada|calz camino|cno|cmo|cmno|cm -camino hondo|ch|c h -camino nuevo|cn|c n -camino viejo|cv|c v +camino hondo|ch|c.h.|c.h|c h +camino nuevo|cn|c.n.|c.n|c n +camino viejo|cv|c.v.|c.v|c v caminito|cmt camping|campg cantera|cantr @@ -74,7 +74,7 @@ paso pasillo|psllo peatonal|peat periferico|perif -plaza|pl|plza|pza|p za +plaza|pl|plza|pza|p.za plazoleta|pzta|plzta|plta plazuela|plzla poblado|pbdo diff --git a/data/dictionaries/es/unit_types.txt b/data/dictionaries/es/unit_types.txt index 08c34d34..ba2d7d8d 100644 --- a/data/dictionaries/es/unit_types.txt +++ b/data/dictionaries/es/unit_types.txt @@ -25,13 +25,13 @@ jardines|jdins|jards mansion|mans mansiones|manses nucleo|ncleo -numero|#|num|no|numr|numro|nro +numero|#|num|no|numr|numro|nro|n.ro officina|off partida|ptda poblado|pbdo|pob|pobl poligono|polig -poligono residencial|polig res|pgres|pg res|pr|p r -poligono industrial|polig ind|pgind|pg ind|pi|p i +poligono residencial|polig res|pgres|pg res|pr|p.r.|p.r|p r +poligono industrial|polig ind|pgind|pg ind|pi|p.i.|p.i|p i portal|prtal|prtl portcio|prtco portillo|ptilo|ptllo diff --git a/data/dictionaries/fr/company_types.txt b/data/dictionaries/fr/company_types.txt index 0b6ee3fd..0089c122 100644 --- a/data/dictionaries/fr/company_types.txt +++ b/data/dictionaries/fr/company_types.txt @@ -1,40 +1,40 @@ association -association sans but lucratif|asbl|a s b l -enterprise individuelle|ei|e i -enterprise unipersonnelle a responsabilite limitee|eurl|e u r l +association sans but lucratif|asbl|a.s.b.l.|a.s.b.l|a s b l +enterprise individuelle|ei|e.i.|e.i|e i +enterprise unipersonnelle a responsabilite limitee|eurl|e.u.r.l.|e.u.r.l|e u r l fondation fondation privee fondation d'utilite publique -fond commun de placement|fcp|f c p +fond commun de placement|fcp|f.c.p.|f.c.p|f c p fonds de placement group|gp groupe|gpe groupement|gpt groups|gps groupes|gpes -groupement d'interet economique|groupement dinteret economique|gie|g i e -raison individuelle|ri|r i -societe anonyme|sa|s a -societe anonyme libanaise|sal|s a l -societe anonyme omanaise generale|saog|s a o g -societe anonyme omanaise close|saoc|s a o c +groupement d'interet economique|groupement dinteret economique|gie|g.i.e.|g.i.e|g i e +raison individuelle|ri|r.i.|r.i|r i +societe anonyme|sa|s.a.|s.a|s a +societe anonyme libanaise|sal|s.a.l.|s.a.l|s a l +societe anonyme omanaise generale|saog|s.a.o.g.|s.a.o.g|s a o g +societe anonyme omanaise close|saoc|s.a.o.c.|s.a.o.c|s a o c societe cooperative|scoop|s coop -societe cooperative a responsabilite limitee|scrl|s c r l -societe cooperative a responsabilite illimitee|scri|s c r i -societe cooperative de production|scop|s c o p -societe d'economie mixte|societe d' economie mixte|societe deconomie mixte|sem|s e m -societe d'investissement a capital fixe|societe dinvestissement a capital fixe|societe d' investissement a capital fixe|sicaf|s i c a f -societe d'investissement a capital variable|societe dinvestissement a capital variable|societe d' investissement a capital variable|sicav|s i c a v +societe cooperative a responsabilite limitee|scrl|s.c.r.l.|s.c.r.l|s c r l +societe cooperative a responsabilite illimitee|scri|s.c.r.i.|s.c.r.i|s c r i +societe cooperative de production|scop|s.c.o.p.|s.c.o.p|s c o p +societe d'economie mixte|societe d' economie mixte|societe deconomie mixte|sem|s.e.m.|s.e.m|s e m +societe d'investissement a capital fixe|societe dinvestissement a capital fixe|societe d' investissement a capital fixe|sicaf|s.i.c.a.f.|s.i.c.a.f|s i c a f +societe d'investissement a capital variable|societe dinvestissement a capital variable|societe d' investissement a capital variable|sicav|s.i.c.a.v.|s.i.c.a.v|s i c a v societe interne societe momentanee -societe a responsabilite limitee|sarl|s a r l -societe en participation|sep|s e p +societe a responsabilite limitee|sarl|s.a.r.l.|s.a.r.l|s a r l +societe en participation|sep|s.e.p.|s.e.p|s e p societe de fait -societe en nom colletif|snc|s n c -societe en commandite|sc|s c -societe en commandite simple|scs|s c s -societe en commandite par actions|sca|s c a -societe par actions simplifiee|sas|s a s -societe privee a responsabilite limitee|sprl|s p r l -societe privee a responsabilite limitee unipersonnelle|sprlu|s p r l u -societe simple|ss|s s \ No newline at end of file +societe en nom colletif|snc|s.n.c.|s.n.c|s n c +societe en commandite|sc|s.c.|s.c|s c +societe en commandite simple|scs|s.c.s.|s.c.s|s c s +societe en commandite par actions|sca|s.c.a.|s.c.a|s c a +societe par actions simplifiee|sas|s.a.s.|s.a.s|s a s +societe privee a responsabilite limitee|sprl|s.p.r.l.|s.p.r.l|s p r l +societe privee a responsabilite limitee unipersonnelle|sprlu|s.p.r.l.u.|s.p.r.l.u|s p r l u +societe simple|ss|s.s.|s.s|s s \ No newline at end of file diff --git a/data/dictionaries/fr/directionals.txt b/data/dictionaries/fr/directionals.txt index 511e71f8..a8de61d0 100644 --- a/data/dictionaries/fr/directionals.txt +++ b/data/dictionaries/fr/directionals.txt @@ -1,4 +1,8 @@ est|e nord|n +nord est|ne|n.e.|n.e|n e +nord ouest|no|n.o.|n.o|n o ouest|o -sud|s \ No newline at end of file +sud|s +sud est|se|s.e.|s.e|s e +sud ouest|so|s.o.|s.o|s o \ No newline at end of file diff --git a/data/dictionaries/fr/personal_titles.txt b/data/dictionaries/fr/personal_titles.txt index cb166a75..7e01812a 100644 --- a/data/dictionaries/fr/personal_titles.txt +++ b/data/dictionaries/fr/personal_titles.txt @@ -28,16 +28,16 @@ monsieur|m|mr pere professeur|prof|pr professeurs|profs|prs -reverende pere|rp|r p -reverende pere jesuit|rpj|r p j -sa saintete|ss|s s +reverende pere|rp|r.p.|r.p|r p +reverende pere jesuit|rpj|r.p.j.|r.p.j|r p j +sa saintete|ss|s.s.|s.s|s s saint|st sainte|ste -son altesse|sa|s a -son altesse imperiale|sai|s a i -son altesse royale|sar|s a r +son altesse|sa|s.a.|s.a|s a +son altesse imperiale|sai|s.a.i.|s.a.i|s a i +son altesse royale|sar|s.a.r.|s.a.r|s a r son eminence|s em -son excellence|se|s e|s exc +son excellence|se|s.e.|s.e|s e|s exc sous lieutenant|souslieutenant|slt|s lt veuve|vve vice amiral|viceamiral|vam|v am \ No newline at end of file diff --git a/data/dictionaries/fr/place_names.txt b/data/dictionaries/fr/place_names.txt index d41579bc..2e1e8e78 100644 --- a/data/dictionaries/fr/place_names.txt +++ b/data/dictionaries/fr/place_names.txt @@ -1,5 +1,5 @@ castel|cst -centre commercial|cc|c c +centre commercial|cc|c.c.|c.c|c c chateau cottage|cott cottages|cott|cotts @@ -13,9 +13,10 @@ parc industriel pont ponts station|sta -zone artisanale|za|z a -zone d'activites|zone dactivites|s a c -zone d'amenagement concerte|zone damenagement concerte|zac|z a c -zone d'amenagement differe|zone damenagement differe|zad|z a d -zone industrielle|zi|z i -zone a urbaniser en priorite|zup|z u p \ No newline at end of file +zone artisanale|za|z.a.|z.a|z a +zone d'activites|zone dactivites|za|z.a.|z.a|z a +zone artisanale commerciale|zac|z.a.c.|z.a.c|z a c +zone d'amenagement concerte|zone damenagement concerte|zac|z.a.c.|z.a.c|z a c +zone d'amenagement differe|zone damenagement differe|zad|z.a.d.|z.a.d|z a d +zone industrielle|zi|z.i.|z.i|z i +zone a urbaniser en priorite|zup|z.u.p.|z.u.p|z u p \ No newline at end of file diff --git a/data/dictionaries/fr/post_office.txt b/data/dictionaries/fr/post_office.txt index 22d57f1f..d6d9e6b2 100644 --- a/data/dictionaries/fr/post_office.txt +++ b/data/dictionaries/fr/post_office.txt @@ -1 +1 @@ -boite postale|bp|b p \ No newline at end of file +boite postale|bp|b.p.|b.p|b p \ No newline at end of file diff --git a/data/dictionaries/fr/street_types.txt b/data/dictionaries/fr/street_types.txt index 621dfcb5..1421cb5e 100644 --- a/data/dictionaries/fr/street_types.txt +++ b/data/dictionaries/fr/street_types.txt @@ -126,7 +126,7 @@ lieu-dit|lieu dit|lieudit|ld|ldit|l dit|l-dit lotissement|lot lotissements|lots mail -maison forestiere|mf|m f +maison forestiere|mf|m.f.|m.f|m f manoir|man marche|mar marches|mars @@ -143,7 +143,7 @@ parcs|prcs parking|pkg parvis|prv passage|pas|psg -passage a niveau|pn|p n +passage a niveau|pn|p.n.|p.n|p n passe|pass passerelle|ple passerelles|ples @@ -194,7 +194,7 @@ roquet|roqt rotonde|rtd|rtnd|rtde|rtnde route|rt|rte routes|rts|rtes -route nationale|rn|r n +route nationale|rn|r.n.|r.n|r n rue|r ruelle|rle ruelles|rles diff --git a/data/dictionaries/it/company_types.txt b/data/dictionaries/it/company_types.txt index 1449975a..507e38c9 100644 --- a/data/dictionaries/it/company_types.txt +++ b/data/dictionaries/it/company_types.txt @@ -1,16 +1,16 @@ -associazione calcio|ac|a c +associazione calcio|ac|a.c.|a.c|a c compagnia|co|comp -cooperativa|coop.va|coop va|coopva -cooperativo|coop.vo|coop vo|coopvo -gruppo di interesse economico|gruppo d'interesse economico|gruppo d' interesse economico|gie|g i e +cooperativa|coop.va|coopva +cooperativo|coop.vo|coopvo +gruppo di interesse economico|gruppo d'interesse economico|gruppo d' interesse economico|gie|g.i.e.|g.i.e|g i e ditta individuale -football club|fc|f c +football club|fc|f.c.|f.c|f c societa|sta -societa cooperativa a responabilita limitata|scrl|s c r l -societa semplice|ss|s s -societa in accomandita semplice|sas|s a s -societa in nome collettivo|snc|s n c -societa per azioni|spa|s p a -societa in accomandita per azioni|sapa|s a p a -societa a garanzia limitata|sagl|s a g l -societa a responsabilita limitata|srl|s r l \ No newline at end of file +societa cooperativa a responabilita limitata|scrl|s.c.r.l.|s.c.r.l|s c r l +societa semplice|ss|s.s.|s.s|s s +societa in accomandita semplice|sas|s.a.s.|s.a.s|s a s +societa in nome collettivo|snc|s.n.c.|s.n.c|s n c +societa per azioni|spa|s.p.a.|s.p.a|s p a +societa in accomandita per azioni|sapa|s.a.p.a.|s.a.p.a|s a p a +societa a garanzia limitata|sagl|s.a.g.l.|s.a.g.l|s a g l +societa a responsabilita limitata|srl|s.r.l.|s.r.l|s r l \ No newline at end of file diff --git a/data/dictionaries/it/directionals.txt b/data/dictionaries/it/directionals.txt index 470e6f47..fcfc4694 100644 --- a/data/dictionaries/it/directionals.txt +++ b/data/dictionaries/it/directionals.txt @@ -2,8 +2,8 @@ nord|n nord est|nordest|ne|n e nord ovest|nordovest|no|n o est|e -occidentale|occid.le|occid le -orientale|orrient.le|orrient le +occidentale|occid.le +orientale|orrient.le ovest|o sud|s sud est|sudest|se|s e diff --git a/data/dictionaries/it/personal_titles.txt b/data/dictionaries/it/personal_titles.txt index 604407b2..484f5faa 100644 --- a/data/dictionaries/it/personal_titles.txt +++ b/data/dictionaries/it/personal_titles.txt @@ -1,8 +1,8 @@ architetto|arch arciprete avvocato|avv -barone|bne|b ne -baronessa|bnssa|bn ssa +barone|bne|b.ne +baronessa|bnssa|bn.ssa capitano|cap caporale caporale maggiore @@ -14,20 +14,20 @@ commendatore|comm console|cons conte contessa -corpo d'armata|corpo darmata|cda|c d a +corpo d'armata|corpo darmata|cda|c.d.a.|c.d.a|c d a curato don donna dottore|dott|dr -dottoressa|dottssa|dott ssa|drssa|dr ssa +dottoressa|dottssa|dott.ssa|drssa|dr.ssa duca duchessa egregio|egr -emissario|emiss.o|emiss o +emissario|emiss.o gentile gentili generale|gen -generali|gen.li|gen li +generali|gen.li grande ufficiale|gr uff giudice imperatore @@ -47,23 +47,23 @@ presidente emerito|pres emerito prete prevosto professore|prof -professoressa|profssa|prof ssa +professoressa|profssa|prof.ssa re regina reverendo|rev -reverendo madre|rm|r m -reverendo padre|rp|r p +reverendo madre|rm|r.m.|r.m|r m +reverendo padre|rp|r.p.|r.p|r p san|s santa|s santo|s -santo padre|sp|s p +santo padre|sp|s.p.|s.p|s p senatore|sen sergente|serg -signora|sigra|sig ra +signora|sigra|sig.ra signore|sig signori|sigg -signorina|signa|sig na -sottotenente|sten|s ten +signorina|signa|sig.na +sottotenente|sten|s.ten tenente|ten tenente colonnello|ten col visconte diff --git a/data/dictionaries/it/place_names.txt b/data/dictionaries/it/place_names.txt index 4a8ea256..f9ae548c 100644 --- a/data/dictionaries/it/place_names.txt +++ b/data/dictionaries/it/place_names.txt @@ -1,141 +1,141 @@ -abbazia|abbadia|badia|abb.a|abb a|b.ia|b ia -abbazie|abbadie|abb.e|abb e -accademia|accad.a|accad a -accademie|accad.e|accad e -acquario|acq.rio|acq rio|acqrio -acquari|acq.ri|acq ri|acqri -acquedotti|acq.ti|acq ti|acqti -acquedotto|acq.to|acq to|acqto -aeroporto|aerop.to|aeropto|aerop to +abbazia|abbadia|badia|abb.a|abba|b.ia|bia +abbazie|abbadie|abb.e|abbe +accademia|accad.a|accada +accademie|accad.e|accade +acquario|acq.rio|acqrio|acqrio +acquari|acq.ri|acqri|acqri +acquedotti|acq.ti|acqti +acquedotto|acq.to|acqto +aeroporto|aerop.to|aeropto agenzia agenzia immobiliare -alberghi|alb.i|alb i -albergo|alb.o|alb o +alberghi|alb.i|albi +albergo|alb.o|albo alimentari -anfiteatri|anfit.i|anfit i -anfiteatro|anfit.o|anfit o -bagno|bgno|b.gno|b gno -bagni|bgni|b.gni|b gni +anfiteatri|anfit.i|anfiti +anfiteatro|anfit.o|anfito +bagno|bgno|b.gno +bagni|bgni|b.gni banca bar -baraccamento|bmento|b.mento|b mento -baraccamenti|bmenti|b.menti|b menti -baracche|bar.che|bar che +baraccamento|bmento|b.mento +baraccamenti|bmenti|b.menti +baracche|bar.che|barche barbiere -basilica|basca|bas.ca|bas ca -basiliche|basche|bas.che|bas che -bettola|bet.a|bet a -bettole|bet.e|bet e +basilica|basca|bas.ca +basiliche|basche|bas.che +bettola|bet.a|beta +bettole|bet.e|bete biblioteca -birreria|birr.a|birr a -birrerie|birr.e|birr e +birreria|birr.a|birra +birrerie|birr.e|birre bottega boutique caffeteria -calzaturificio|calzat.cio|calzat cio +calzaturificio|calzat.cio|calzatcio calzolaio -campeggio|camp.gio|camp gio|campgio -campeggi|camp.gi|camp gi|campgi -cappella|cap.la|cap la|capla +campeggio|camp.gio|campgio +campeggi|camp.gi|campgi +cappella|cap.la|capla cappelleria -cappelletta|cap.ta|cap ta|capta -carabinieri|carab.ri|carabri|carb ri -carcere|carc.e|carc e -carenaggio|car.ggio|car ggio|carggio +cappelletta|cap.ta|capta +carabinieri|carab.ri|carabri +carcere|carc.e|carce +carenaggio|car.ggio|carggio cartiera cartoleria casa -casale|c.ale|c ale -casaletto|c.letto|c letto -caserma|cas.ma|cas ma|casma -casetta|c.ta|c ta -casette|c.te|c te +casale|c.ale|cale +casaletto|c.letto|cletto +caserma|cas.ma|casma +casetta|c.ta|cta +casette|c.te|cte casino -casolare|c.lare|c lare -casone|c.one|c one -casotto|cas.to|cas to -cassina|cass.a|cass a -castello|cast.o|cast o -catacomba|cat.ba|cat ba|catba -catacombe|cat.be|cat be|catbe -cattedrale|catt.le|catt le -chalet|ch.t|ch t -chiesa|ch.sa|ch sa -chiesetta|ch.etta|ch etta -cittadella|c.della +casolare|c.lare|clare +casone|c.one|cone +casotto|cas.to|casto +cassina|cass.a|cassa +castello|cast.o|casto +catacomba|cat.ba|cat ba +catacombe|cat.be|cat be +cattedrale|catt.le|cattle +chalet|ch.t|cht +chiesa|ch.sa|chsa +chiesetta|ch.etta|chetta +cittadella|c.della|cdella club -collegio|coll.io|coll io -conservatorio|cons.rio|cons rio|consrio -convento|conv.to|conv to|convto +collegio|coll.io|collio +conservatorio|cons.rio|consrio +convento|conv.to|convto diga -dissalazione|dissal.ne|dissal ne -distilleria|distill.a|distill a +dissalazione|dissal.ne|dissalne +distilleria|distill.a|distilla drogheria enoteca -ergastolo|erg.lo|erg lo -fabbrica|fabbr.a|fabbr a -faro|f.ro|f ro +ergastolo|erg.lo|erglo +fabbrica|fabbr.a|fabbra +faro|f.ro|fro farmacia -fattoria|fatt.a|fatt a +fattoria|fatt.a|fatta ferramenta -ferriera|ferr.ra|ferr ra -ferrovia|ferr.a|ferr a|fv|f v -ferroviaria|ferr.ria|ferr ria -ferroviario|ferr.rio|ferr rio +ferriera|ferr.ra|ferrra +ferrovia|ferr.a|fv|f.v.|f.v|f v +ferroviaria|ferr.ria|ferrria +ferroviario|ferr.rio|ferrrio fiera fioraio -fonderia|fond.a|fond a -fontana|font.na|font na -forni|f.ni|f ni -forno|f.no|f no +fonderia|fond.a +fontana|font.na|fontna +forni|f.ni|fni +forno|f.no|fno foro forte -fortezza|fort.za|fort za +fortezza|fort.za|fortza forti -fortificazione|fort.one|fort one -fortilizio|fort.zio|fort zio +fortificazione|fort.one|fortone +fortilizio|fort.zio|fortzio fratelli galleria gelateria -gendarmeria|gend.ria|gend ria +gendarmeria|gend.ria|gendria giolleria giornalaio hotel -impianto|imp.to|imp to -ippodromo|ip.dromo|ip dromo +impianto|imp.to|impto +ippodromo|ip.dromo|ipdromo istituto|ist kebab -laboratorio|lab.o|lab o +laboratorio|lab.o|labo latteria -lavanderia|lavand.a|lavand a +lavanderia|lavand.a|lavanda lavanderia a gettone libreria locanda|loc.a -lunetta|l.etta|l etta +lunetta|l.etta macelleria -madonnella|mad.la|mad la -madonnetta|mad.ta|mad ta -magazzini|magaz.i|magaz i -magazzino|magaz.o|magaz o +madonnella|mad.la|madla +madonnetta|mad.ta|madta +magazzini|magaz.i|magazi +magazzino|magaz.o|magazo masseria|mass.a -ministero|min.ro|min ro +ministero|min.ro|minro molino -monastero|monast.o|monast o -monumento|monum.o|monum o +monastero|monast.o|monasto +monumento|monum.o|monumo mulino museo negozio -obelisco|ob.sco|ob sco|obsco +obelisco|ob.sco|obsco orologiaio -ospedale|osp.le|osp le|osple +ospedale|osp.le|osple ostello -osteria|ost.a|ost a +osteria|ost.a|osta ottica -padiglinoe|pad.ne|pad ne|padne +padiglinoe|pad.ne|padne panetteria panificio parcheggio|pach.o -parco|p.rco|p rco|prco +parco|p.rco|prco parrucchiere pasticceria pelletteria diff --git a/data/dictionaries/it/post_office.txt b/data/dictionaries/it/post_office.txt index 6fcc90c1..c7c72c08 100644 --- a/data/dictionaries/it/post_office.txt +++ b/data/dictionaries/it/post_office.txt @@ -1 +1 @@ -casella postale|cp|c p \ No newline at end of file +casella postale|cp|c.p.|c.p|c p \ No newline at end of file diff --git a/data/dictionaries/it/qualifiers.txt b/data/dictionaries/it/qualifiers.txt index 05ec8105..a8bc1dc1 100644 --- a/data/dictionaries/it/qualifiers.txt +++ b/data/dictionaries/it/qualifiers.txt @@ -1,16 +1,16 @@ -borgata|borg.a|borga|borg a -borgate|borg.e|borge|borg e -borghi|b.ghi|bghi|b ghi -borgo|b.go|bgo|b go -blocchi|bl.i|bl i -blocco|bl.o|bl o -colonia|col.a|col a +borgata|borg.a|borga +borgate|borg.e|borge +borghi|b.ghi|bghi +borgo|b.go|bgo +blocchi|bl.i|bli +blocco|bl.o|blo +colonia|col.a|cola fermata -fortino|fort.no|fort no +fortino|fort.no|fortno isola isole isoletta isolette isoletto isoletti -palazzi|pal.zi \ No newline at end of file +palazzi|pal.zi|palzi \ No newline at end of file diff --git a/data/dictionaries/it/stopwords.txt b/data/dictionaries/it/stopwords.txt index abaf7826..756c3ef7 100644 --- a/data/dictionaries/it/stopwords.txt +++ b/data/dictionaries/it/stopwords.txt @@ -17,11 +17,11 @@ dall' dei degli del -della|d.la|d la -delle|d.le|d le -dello|d.lo|d lo +della|d.la|dla +delle|d.le|dle +dello|d.lo|dlo dell' -dentro|d.tro|d tro +dentro|d.tro|dtro di d' fuori diff --git a/data/dictionaries/it/street_types.txt b/data/dictionaries/it/street_types.txt index 8afc6327..ecbc4c81 100644 --- a/data/dictionaries/it/street_types.txt +++ b/data/dictionaries/it/street_types.txt @@ -1,55 +1,55 @@ -accesso|acc.so|acc so|accso -autostrada|aut.da|autda|aut da -audostradale|aut.sle|autsle|aut sle|aut.ale|autale|aut ale -autostrade|aut.de|autde|aut de -bretella|br.lla|brlla|br lla -bretelle|br.lle|brlle|br lle -calle|c.le|c le|cle -campo|c.po|c po|cpo -cavalcavia|cavalc.a|cavalc a -chiusa|c.usa|c usa -circonvallazione|circonv.e|circonv e -contrada|contr.a|contr a -corso|c.so|c so|cso -corte|c.te|c te|cte -cortile|c.ile|c ile -ferrata|f.rata|f rata -fondamenta|f.ta|f ta|fta -forca|f.ca|f ca -forcella|forc.la|forc la -frazione|fraz.e|fraz e -frazioni|fraz.i|fraz i -galleria|gall.a|gall a -gallerie|gall.e|gall e -genna|g.na|g na -genne|g.ne|g ne -largo|l.go|l go|lgo -localita|loc|loc.ta|loc ta -piazza|p.za|p.zza|p za|p zza|pza|pzza -piazzale|p.le|p le|ple -piazzetta|p.ta|p ta|pta -ponte|p.te|p te|pte -porta|p.ta|p ta|pta +accesso|acc.so|accso +autostrada|aut.da|autda +audostradale|aut.sle|autsle|aut sle|aut.ale|autale +autostrade|aut.de|autde +bretella|br.lla|brlla +bretelle|br.lle|brlle +calle|c.le|cle +campo|c.po|cpo +cavalcavia|cavalc.a|cavalca +chiusa|c.usa|cusa +circonvallazione|circonv.e|circonve +contrada|contr.a|contra +corso|c.so|cso +corte|c.te|cte +cortile|c.ile|ctile +ferrata|f.rata|frata +fondamenta|f.ta|fta +forca|f.ca|fca +forcella|forc.la|forcla +frazione|fraz.e|fraze +frazioni|fraz.i|frazi +galleria|gall.a|galla +gallerie|gall.e|galle +genna|g.na|gna +genne|g.ne|gne +largo|l.go|lgo +localita|loc|loc.ta|locta +piazza|p.za|p.zza|pza|pzza +piazzale|p.le|ple +piazzetta|p.ta|pta +ponte|p.te|pte +porta|p.ta|pta provinciale -salizada|s.da|s da|sda +salizada|s.da|sda stazione|staz -strada|s|str.a|str a|stra -strada comunale|sc|s c -strada provinciale|sp|s p|str provinciale|s provinciale -strada regionale|sr|s r -strada statale|ss|s s -strada vicinale|sv|s v -stretta|str.ta|str ta|strta -strette|str.te|str te|strte -stretti|str.ti|str ti|strti -stretto|str.to|str to|strto +strada|s|str.a|stra +strada comunale|sc|s.c.|s.c|s c +strada provinciale|sp|s.p.|s.p|s p|str provinciale|s provinciale +strada regionale|sr|s.r.|s.r|s r +strada statale|ss|s.s.|s.s|s s +strada vicinale|sv|s.v.|s.v|s v +stretta|str.ta|strta +strette|str.te|strte +stretti|str.ti|strti +stretto|str.to|strto superstrada tangenziale traversa via|v -via comunale|vc|v c -via provincale|vp|v p|v provinciale -via statale|vs|v s -viale|v.le|v le|vle -vicolo|v.lo|v lo|vlo +via comunale|vc|v.c.|v.c|v c +via provincale|vp|v.p.|v.p|v p|v provinciale +via statale|vs|v.s.|v.s|v s +viale|v.le|vle +vicolo|v.lo|vlo vico \ No newline at end of file diff --git a/data/dictionaries/it/synonyms.txt b/data/dictionaries/it/synonyms.txt index 6f0b9c0b..23fdb970 100644 --- a/data/dictionaries/it/synonyms.txt +++ b/data/dictionaries/it/synonyms.txt @@ -1,115 +1,115 @@ agosto|ago -altezza|alt.za|alt za|altza -altezze|alt.ze|alt ze|altze -antica|ant.a|ant a -antiche|ant.e|ant e -antichi|ant.i|ant i -antico|ant.o|ant o +altezza|alt.za|altza +altezze|alt.ze|altze +antica|ant.a|anta +antiche|ant.e|ante +antichi|ant.i|anti +antico|ant.o|anto aprile|apr -arcipelaghi|arcip.ghi|arcip ghi|arcipghi -arcipelagho|arcip.gho|arcip gho|arcipgho -badde|b.de|b de -bassa|b.sa|b sa|bsa -bassi|b.si|b si|bsi -basso|b.so|b so|bso -basse|b.se|b se|bse -bealera|b.ra|b ra|bra -bealere|b.re|b re|bre -calvario|calv.o|calv o -canale|can.le|can le -canali|can.li|can li -caverna|cav.na|cav na -caverne|cav.ne|cav ne -centrale|centr.e|centr e -colla|c.la -colle|c.le -colletta|c.tta|c tta -collina|cll.a|cll a -colline|cll.e|cll e -collo|c.llo -cozzi|c.zi|c zi|czi -cozzo|c.zo|c zo|czo -commerciale|comm.le -comunale|com.le -comunita|com.ta|com ta +arcipelaghi|arcip.ghi|arcipghi +arcipelagho|arcip.gho|arcipgho +badde|b.de|bde +bassa|b.sa|bsa +bassi|b.si|bsi +basso|b.so|bso +basse|b.se|bse +bealera|b.ra|bra +bealere|b.re|bre +calvario|calv.o|calvo +canale|can.le|canle +canali|can.li|canli +caverna|cav.na|cavna +caverne|cav.ne|cavne +centrale|centr.e|ctrle +colla|c.la|cla +colle|c.le|cle +colletta|c.tta|ctta +collina|cll.a|clla +colline|cll.e|clle +collo|c.llo|cllo +cozzi|c.zi|czi +cozzo|c.zo|czo +commerciale|comm.le|commle +comunale|com.le|comle +comunita|com.ta|comta decembre|dec dicembre|dic -distretto|distr.o|distr o -diversivo|div.vo|div vo -fanteria|fant.a|fant a +distretto|distr.o|distro +diversivo|div.vo|divvo +fanteria|fant.a|fanta febbraio|feb|fev -fontanile|font.ile|font ile -fonte|f.te|f te -fosso|f.so|f so|fso -fossa|f.sa|f sa|fsa -fratelli|flli|f lli +fontanile|font.ile|fontile +fonte|f.te|fte +fosso|f.so|fso +fossa|f.sa|fsa +fratelli|f.lli|flli gennaio|gen -ghiacciai|gh.i|gh i -ghiacciaio|gh.io|gh io -giardino|g.no|g no -giardini|g.ni|g ni +ghiacciai|gh.i|ghi +ghiacciaio|gh.io|ghio +giardino|g.no|gno +giardini|g.ni|gni giugno|giug giulio|giul gran|gr -grande|gr.de|gr de -grandi|gr.di|gr di -grossa|gr.sa|gr sa -grosse|gr.se|gr se -grossi|gr.si|gr si -grosso|gr.so|gr so -grotta|gr.ta|gr ta -gortte|gr.te|gr te -gruppi|gr.pi|gr pi -gruppo|gr.po|gr po +grande|gr.de|grde +grandi|gr.di|grdi +grossa|gr.sa|grsa +grosse|gr.se|grse +grossi|gr.si|grsi +grosso|gr.so|grso +grotta|gr.ta|grta +gortte|gr.te|grte +gruppi|gr.pi|grpi +gruppo|gr.po|grpo incoronata|incor.ta -industria|ind.ia|ind ia -industriale|ind.ale|ind ale -industriali|ind.ali|ind ali -inferiore|inf.e|inf e -inferiori|inf.i|inf i -internazionale|intern.le|intern le -internazionali|intern.li|intern li -interne|int.ne|int ne -interni|int.ni|int ni -interno|int.no|int no -laghetto|lag.to|lag to +industria|ind.ia +industriale|ind.ale|indale +industriali|ind.ali|indali +inferiore|inf.e|infe +inferiori|inf.i|infi +internazionale|intern.le +internazionali|intern.li +interne|int.ne|intne +interni|int.ni|intni +interno|int.no|intno +laghetto|lag.to|lagto lago|l -laguna|lag.a|lag a -lunghi|l.ghi|l ghi -lungo|l.go|l go +laguna|lag.a +lunghi|l.ghi|lghi +lungo|l.go|lgo maggio|mag -maddalena|madd.na|madd na -madonna|mad.na|mad na -madonne|mad.ne|mad ne -maggiore|magg.re|magg re -maggiori|magg.ri|magg ri -mandamento|mand.to|mand to -mannu|m.nu|m nu -marinella|marin.la|marin la -marino|m.ino|m ino -marittimo|maritt.o|maritt o +maddalena|madd.na|maddna +madonna|mad.na|madna +madonne|mad.ne|madne +maggiore|magg.re|maggre +maggiori|magg.ri|maggri +mandamento|mand.to|mandto +mannu|m.nu|mnu +marinella|marin.la|marinla +marino|m.ino|mrno +marittimo|maritt.o marzo|mar media|med.a -mediterraneo|medit.eo|medit eo -metropolitana|metr.na|metr na|metro -militare|mil.re|mil re -muri|m.ri|m ri -muro|m.ro|m ro -navigli|nav.i|nav i -naviglio|nav.o|nav o -nazionale|naz.le|naz le -nazionali|naz.li|naz li -nostra|n.ra|n ra -nostro|n.ro|n ro +mediterraneo|medit.eo +metropolitana|metr.na|metrna|metro +militare|mil.re|milre +muri|m.ri|mri +muro|m.ro|mro +navigli|nav.i +naviglio|nav.o +nazionale|naz.le|nazle +nazionali|naz.li|nazli +nostra|n.ra|nra +nostro|n.ro|nro novembre|nov -nuova|n.va|n va -nuove|n.ve|n ve -nuovi|n.vi|n vi -nuovo|n.vo|n vo +nuova|n.va|nva +nuove|n.ve|nve +nuovi|n.vi|nvi +nuovo|n.vo|nvo ottobre|ott penisola|penis.a -piccola|p.cola|p cola -piccole|p.cole|p cole -piccoli|p.coli|p coli -piccolo|p.colo|p colo +piccola|p.cola|pcola +piccole|p.cole|pcole +piccoli|p.coli|pcoli +piccolo|p.colo|pcolo settembre|set \ No newline at end of file diff --git a/data/dictionaries/it/unit_types.txt b/data/dictionaries/it/unit_types.txt index f545329b..0048c50b 100644 --- a/data/dictionaries/it/unit_types.txt +++ b/data/dictionaries/it/unit_types.txt @@ -1,13 +1,13 @@ -blocco|bl.o|blo|bl o -blocchi|bl.i|bli|bl i -cantina|c.tina|c tina|ctina +blocco|bl.o|blo|blo +blocchi|bl.i|bli|bli +cantina|c.tina|ctina casa|c -casetto|c.etto|c etto -dipartimento|dipart.o|dipart o -edificio|edif.o|edif o -garage|g.ge|g ge -giardino|g.no|g no +casetto|c.etto +dipartimento|dipart.o|dpmto +edificio|edif.o|edifo +garage|g.ge|gge +giardino|g.no|gno numero|no|n.o|n o|n -officina|off.a|off a|off -palazzo|pal.zo|pal zo|palzo -piano|p.o|p o|p.no|p no|pno \ No newline at end of file +officina|off.a|off +palazzo|pal.zo|palzo +piano|p.o|p.no|pno \ No newline at end of file diff --git a/data/dictionaries/nl/directionals.txt b/data/dictionaries/nl/directionals.txt index f71d88ed..9371c67f 100644 --- a/data/dictionaries/nl/directionals.txt +++ b/data/dictionaries/nl/directionals.txt @@ -1,8 +1,8 @@ noord|n -noordoost|noord oost|noordoosten|noord oosten|no|n o -noordwest|noord west|noordwesten|noord westen|nw|n w +noordoost|noord oost|noordoosten|noord oosten|no|n.o.|n.o|n o +noordwest|noord west|noordwesten|noord westen|nw|n.w.|n.w|n w oost|o west|w -zuidoost|zuid oost|zuidoosten|zuid oosten|zo|z o -zuidwest|zuidwesten|zw|z w +zuidoost|zuid oost|zuidoosten|zuid oosten|zo|z.o.|z.o|z o +zuidwest|zuidwesten|zw|z.w.|z.w|z w zuid|z \ No newline at end of file diff --git a/data/dictionaries/nl/qualifiers.txt b/data/dictionaries/nl/qualifiers.txt index 402afcd0..ce2de416 100644 --- a/data/dictionaries/nl/qualifiers.txt +++ b/data/dictionaries/nl/qualifiers.txt @@ -1,4 +1,4 @@ -noordzijde|nz|n z -oostzijde|oz|o z -westzijde|wz|w z -zuidzijde|zz|z z \ No newline at end of file +noordzijde|nz|n.z.|n.z|n z +oostzijde|oz|o.z.|o.z|o z +westzijde|wz|w.z.|w.z|w z +zuidzijde|zz|z.z.|z.z|z z \ No newline at end of file diff --git a/data/dictionaries/pt/company_types.txt b/data/dictionaries/pt/company_types.txt index 90a694ae..3d2b048f 100644 --- a/data/dictionaries/pt/company_types.txt +++ b/data/dictionaries/pt/company_types.txt @@ -1,31 +1,31 @@ -agrupamento complementar de empresas|ace|a c e -associacao desportiva|ad|a d +agrupamento complementar de empresas|ace|a.c.e.|a.c.e|a c e +associacao desportiva|ad|a.d.|a.d|a d associacao em sentido estrito -clube desportivo|cd|c d +clube desportivo|cd|c.d.|c.d|c d companhia|cia|comp -companhia anonima|ca|c a -companhia de responsabilidao limitada|crl|c r l -empresa individual|ei|e i|firma individual -empresa individual de responsabilidade limitada|eireli|e i r e l i -empresa publica|ep|e p +companhia anonima|ca|c.a.|c.a|c a +companhia de responsabilidao limitada|crl|c.r.l.|c.r.l|c r l +empresa individual|ei|e i|e.i.|e.i|firma individual +empresa individual de responsabilidade limitada|eireli|e.i.r.e.l.i.|e.i.r.e.l.i|e i r e l i +empresa publica|ep|e.p.|e.p|e p fundacao privada fundacao publica -futebol clube|fc|f c -limitada|ltda|lt da -organizacao nao governamental|ong|o n g -organizacao da sociedade civil de interesse publico|oscip|o s c i p -sociedade aberta|sa|s a -sociedade anonima|sa|s a|s / a -sociedade anonima de responsabilidade limitada|sarl|s a r l -sociedade anonima desportiva|sad|s a d -sociedade em nome coletivo|snc|s n c -sociedade em conta de participacao|scp|s c p -sociedade em comandita simples|scs|s c s +futebol clube|fc|f.c.|f.c|f c +limitada|ltda|lt.da|lda|l.da| +organizacao nao governamental|ong|o.n.g.|o.n.g|o n g +organizacao da sociedade civil de interesse publico|oscip|o.s.c.i.p.|o.s.c.i.p|o s c i p +sociedade aberta|sa|s.a.|s.a|s a +sociedade anonima|sa|s.a.|s.a|s a|s / a +sociedade anonima de responsabilidade limitada|sarl|s.a.r.l.|s.a.r.l|s a r l +sociedade anonima desportiva|sad|s.a.d.|s.a.d|s a d +sociedade em nome coletivo|snc|s.n.c.|s.n.c|s n c +sociedade em conta de participacao|scp|s.c.p.|s.c.p|s c p +sociedade em comandita simples|scs|s.c.s.|s.c.s|s c s sociedade em comum -sociedade fechada|sf|s f -sociedade gestora de participacoes socials|sgps|s g p s -sociedade limitada|sociedade por quotas limitada|limitada|lda|sl|s l -sociedade por acoes|sa|s a -sociedade simples|pllc|p l l c -sport clube|sc|s c -unipessoal lda|ulda|u l d a \ No newline at end of file +sociedade fechada|sf|s.f.|s.f|s f +sociedade gestora de participacoes socials|sgps|s.g.p.s.|s.g.p.s|s g p s +sociedade limitada|sociedade por quotas limitada|sl|s.l.|s.l|s l|s limitada|s ltda|s lt.da|s lda|s l.da +sociedade por acoes|sa|s.a.|s.a|s a +sociedade simples|pllc|p.l.l.c.|p.l.l.c|p l l c +sport clube|sc|s.c.|s.c|s c +unipessoal lda|ulda|u.l.d.a.|u.l.d.a|u l d a \ No newline at end of file diff --git a/data/dictionaries/pt/directionals.txt b/data/dictionaries/pt/directionals.txt index ad8e84f3..51c4b196 100644 --- a/data/dictionaries/pt/directionals.txt +++ b/data/dictionaries/pt/directionals.txt @@ -1,8 +1,8 @@ este|e norte|n -noredeste|ne|n e -noroeste|no|n o +noredeste|ne|n.e.|n.e|n e +noroeste|no|n.o.|n.o|n o oeste|o -sudeste|se|s e -sudoeste|so|s o +sudeste|se|s.e.|s.e|s e +sudoeste|so|s.o.|s.o|s o sul|s \ No newline at end of file diff --git a/data/dictionaries/pt/no_number.txt b/data/dictionaries/pt/no_number.txt index 12220985..1e856294 100644 --- a/data/dictionaries/pt/no_number.txt +++ b/data/dictionaries/pt/no_number.txt @@ -1 +1 @@ -sem numero|sn|s / n|s n \ No newline at end of file +sem numero|sn|s.n.|s.n|s / n|s n \ No newline at end of file diff --git a/data/dictionaries/pt/personal_suffixes.txt b/data/dictionaries/pt/personal_suffixes.txt index c7364136..c0c1480a 100644 --- a/data/dictionaries/pt/personal_suffixes.txt +++ b/data/dictionaries/pt/personal_suffixes.txt @@ -1,3 +1,3 @@ -filho|f o +filho|f.o junior|jr neto \ No newline at end of file diff --git a/data/dictionaries/pt/personal_titles.txt b/data/dictionaries/pt/personal_titles.txt index d68f31d1..4c2bf2d2 100644 --- a/data/dictionaries/pt/personal_titles.txt +++ b/data/dictionaries/pt/personal_titles.txt @@ -1,10 +1,10 @@ -advogado|advo|adv o +advogado|advo|adv.o alferes|alf almirante|alm -arcebispo|arco|arc o +arcebispo|arco|arc.o barao|bar baronesa -bispo|bpo|b po +bispo|bpo|b.po brigadeiro|brg|brig cacique|cac capitao|cap|capt @@ -14,7 +14,7 @@ capitao de mar e guerra|cap m g|cap mg capitao tenente|cap ten|cap tte cardeal|card comandante|cmdt -comendador|comend|cmdr|comor|com or +comendador|comend|cmdr|comor|com.or conde|cde condessa|cdessa conselheiro|cons @@ -22,30 +22,30 @@ coronel|cor|cel contra almirante|c alm deputado|dep desembargador|des -desembargadora|desa|des a +desembargadora|desa|des.a diacono|diac diacona|diaca dom|d -dona|da +dona|da|d.a doutor|dr -doutora|dra +doutora|dra|d.ra duque|dq duquesa|dqsa|dqa embaixador|emb -eminencia|ema|em a -eminentissimo|emmo|em mo -enfermeira|enfa|enf a +eminencia|ema|em.a +eminentissimo|emmo|em.mo +enfermeira|enfa|enf.a enfermeiro|enfo|enf engenheira|enga engenheiro|eng|engo -excelencia|exa|ex a -excelentissima|exma|ex ma -excelentissimo|exmo|ex mo +excelencia|exa|ex.a +excelentissima|exma|ex.ma +excelentissimo|exmo|ex.mo frei|fr -general|gen|gal|g al -governador|gov|govdor|govd or|gov dor -ilustrissima|ilma|il ma -ilustrissimo|ilmo|il mo +general|gen|gal|g.al +governador|gov|govdor|govd.or|gov.dor +ilustrissima|ilma|il.ma +ilustrissimo|ilmo|il.mo infante|inf irma|ima|im irmao|imo|im @@ -54,18 +54,18 @@ maestra|mta|mtra|mstra maestro|mto|mtro|mstro major|maj major brigadeiro|maj brig -marechal|mar|mal|m al +marechal|mar|mal|m.al marques|mq|marq marquesa|mqsa|marqa madre|me medico|med -mestra|ma|m a -mestre|me|m e +mestra|ma|m.a +mestre|me|m.e ministerio|min monsenhor|mons -nossa senhora|nsa|ns a|n sa|n s a -nosso senhor|ns|n s -padre|pe|p e +nossa senhora|nsa|ns a|n sa|n.s.a.|n.s.a +nosso senhor|ns|n.s.|n.s|n s +padre|pe|p.e pastor|pr prefeito|pref presidente|presid|pres @@ -73,29 +73,29 @@ professor|prof professora|profa professoras|profas professores|profs -reverenda|revda|rev da -reverendo|revdo|rev do -reverendissima|revma|rev ma -reverendissimo|revmo|rev mo +reverenda|revda|rev.da +reverendo|revdo|rev.do +reverendissima|revma|rev.ma +reverendissimo|revmo|rev.mo santa|sta santo|st|sto sao|s sargento|sarg|sgto|sargto -sargento ajudante|sarg ajte|sarg aj te -secretaria|sec a|secr a|secra +sargento ajudante|sarg ajte|sarg aj.te +secretaria|sec.a|secr.a|secra secretario|sec|secr senador|sen -senhor|s|sr|s r -senhores|srs|s rs -senhora|sa|sra|sr a|s ra -senhoras|sras|s ras -senhorita|srta|sr ta -senhoritas|srtas|sr tas -tenente|ten|tte|t te -tenente coronel|ten c el|ten cel|tte cel|t te c el +senhor|s|sr|s.r +senhores|srs|s.rs +senhora|sa|sra|sr.a|s.ra +senhoras|sras|s.ras +senhorita|srta|sr.ta +senhoritas|srtas|sr.tas +tenente|ten|tte|t.te +tenente coronel|ten c.el|ten cel|tte cel|t.te c.el tesoureiro|tes vereador|ver vice almirante|v alm -vigario|vig|vigo|vig o -visconde|visc|vde|v de -viscondesa|vdessa|v dessa \ No newline at end of file +vigario|vig|vigo|vig.o +visconde|visc|vde|v.de +viscondesa|vdessa|v.dessa \ No newline at end of file diff --git a/data/dictionaries/pt/place_names.txt b/data/dictionaries/pt/place_names.txt index 44df99e5..8b23e48f 100644 --- a/data/dictionaries/pt/place_names.txt +++ b/data/dictionaries/pt/place_names.txt @@ -1,27 +1,27 @@ academia|acad associacao|ass -arquitecto|arq|arqo -bairro|b|bo|br +arquitecto|arq|arqo|arq.o +bairro|b|bo|br|b.o bombeiros voluntarios|bv|b v -camara municipal|cm|c m +camara municipal|cm|c.m.|c.m|c m casal|csl castelo|cast -centro comercial|cc|c c -ciclo do ensino basico|ceb|c e b -escola basica|eb|e b -estacao de tratamento de aguas residuais|etar|e t a r -fonte +centro comercial|cc|c.c.|c.c|c c +ciclo do ensino basico|ceb|c.e.b.|c.e.b|c e b +escola basica|eb|e.b.|e.b|e b +estacao de tratamento de aguas residuais|etar|e.t.a.r.|e.t.a.r|e t a r +fonte|f.te|fte galeria|gal -guarda nacional republicana|gnr|g n r +guarda nacional republicana|gnr|g.n.r.|g.n.r|g n r instituto|inst loja lojas parque|pq|prq|parq pavilhao|pav -policia de seguranca publica|psp|p s p -policia judiciaria|pj|p j -ponte|pont|pte|pnte -porto|pto +policia de seguranca publica|psp|p.s.p.|p.s.p|p s p +policia judiciaria|pj|p.j.|p.j|p j +ponte|pont|pte|p.te|pnte +porto|pto|p.to rodoviaria|rdv seminario|sem|semin universidade|univ|univers \ No newline at end of file diff --git a/data/dictionaries/pt/qualifiers.txt b/data/dictionaries/pt/qualifiers.txt index 60c36816..449cec07 100644 --- a/data/dictionaries/pt/qualifiers.txt +++ b/data/dictionaries/pt/qualifiers.txt @@ -3,5 +3,5 @@ distrito|dtto|dist divisao|div quadra|q|qd secao|sec -superquadra|super quadra|sq|s q +superquadra|super quadra|sq|s.q.|s.q|s q zona|zn \ No newline at end of file diff --git a/data/dictionaries/pt/stopwords.txt b/data/dictionaries/pt/stopwords.txt index 1a9e9fee..cb8d0565 100644 --- a/data/dictionaries/pt/stopwords.txt +++ b/data/dictionaries/pt/stopwords.txt @@ -22,7 +22,7 @@ na nas no nos -nossa|n|na +nossa|n|na|n.a nosso|n o os diff --git a/data/dictionaries/pt/synonyms.txt b/data/dictionaries/pt/synonyms.txt index 7f1ac029..28664a1e 100644 --- a/data/dictionaries/pt/synonyms.txt +++ b/data/dictionaries/pt/synonyms.txt @@ -1,18 +1,18 @@ abril|abr agosto|ago -dezembro|dezbro|dezbr o|dezb ro|dez bro|dez|10bro|10 bro|10b ro|10br o +dezembro|dezbro|dezbr.o|dezb.ro|dez.bro|dez|10bro|10.bro|10b.ro|10br.o educacao|educ -fevereiro|fev +fevereiro|fev|fevro|fev.ro informacao|inf|inform -janeiro|janero|janro|janr o|jan ro|jan +janeiro|janero|janro|janr.o|jan.ro|jan junho|jun julho|jul março|mar maio|mai monte|mnt|mte|mnte -liberdade|liberde|liberd e -novembro|9bro|9 bro|9b ro|9br o|nov -outubro|8bro|8 bro|8b ro|8br o|out -ribeira|ra|r a|riba|ribra|ribr a -ribeiro|ro|r o|ribo|ribro|ribr o -setembro|7bro|7 bro|7b ro|7br o|set \ No newline at end of file +liberdade|liberde|liberd.e +novembro|9bro|9.bro|9b.ro|9br.o|nov +outubro|8bro|8.bro|8b.ro|8br.o|out +ribeira|ra|r.a|riba|ribra|ribr.a +ribeiro|ro|r.o|ribo|ribro|ribr.o +setembro|7bro|7.bro|7b.ro|7br.o|set \ No newline at end of file diff --git a/data/dictionaries/pt/toponyms.txt b/data/dictionaries/pt/toponyms.txt index e9fd7568..9dbdfb8d 100644 --- a/data/dictionaries/pt/toponyms.txt +++ b/data/dictionaries/pt/toponyms.txt @@ -1 +1 @@ -lisboa|lxa|lx a|lex a|lexa \ No newline at end of file +lisboa|lxa|lx.a|lex.a|lexa \ No newline at end of file diff --git a/data/dictionaries/pt/unit_types.txt b/data/dictionaries/pt/unit_types.txt index f26ef6be..ba3dc3f2 100644 --- a/data/dictionaries/pt/unit_types.txt +++ b/data/dictionaries/pt/unit_types.txt @@ -1,5 +1,5 @@ andar -apartamento|apt|apto +apartamento|apt|apto|apt.o batalhao|btl cave|cv condominio