From 2bae8075b00fecfd92494e069b8f128552bf5e48 Mon Sep 17 00:00:00 2001 From: jeffrey04 Date: Wed, 28 Sep 2016 18:41:15 +0800 Subject: [PATCH 01/12] initial commit of malay words --- resources/dictionaries/en/company_types.txt | 2 +- .../dictionaries/ms/ambiguous_expansion.txt | 6 + resources/dictionaries/ms/building_types.txt | 8 + resources/dictionaries/ms/company_types.txt | 6 + resources/dictionaries/ms/directionals.txt | 18 +-- resources/dictionaries/ms/given_names.txt | 1 + resources/dictionaries/ms/level_types.txt | 1 + resources/dictionaries/ms/personal_titles.txt | 95 +++++++++++- resources/dictionaries/ms/place_names.txt | 139 +++++++++++++++++- resources/dictionaries/ms/post_office.txt | 2 + resources/dictionaries/ms/qualifiers.txt | 5 +- resources/dictionaries/ms/street_types.txt | 33 ++++- resources/dictionaries/ms/synonyms.txt | 25 ++++ 13 files changed, 320 insertions(+), 21 deletions(-) create mode 100644 resources/dictionaries/ms/ambiguous_expansion.txt create mode 100644 resources/dictionaries/ms/building_types.txt create mode 100644 resources/dictionaries/ms/company_types.txt create mode 100644 resources/dictionaries/ms/given_names.txt create mode 100644 resources/dictionaries/ms/level_types.txt create mode 100644 resources/dictionaries/ms/post_office.txt create mode 100644 resources/dictionaries/ms/synonyms.txt diff --git a/resources/dictionaries/en/company_types.txt b/resources/dictionaries/en/company_types.txt index 3af64e20..b6396677 100644 --- a/resources/dictionaries/en/company_types.txt +++ b/resources/dictionaries/en/company_types.txt @@ -39,7 +39,7 @@ national trust and savings association|national trust & savings association|nt & no liability|nl|n l nonprofit|non profit open ended investment company|oeic|o e i c -private limited company|pvt ltd +private limited company|pvt ltd|pte ltd professional limited liability company|pllc|p l l c professional service corporation|psc|p s c proprietary|pty diff --git a/resources/dictionaries/ms/ambiguous_expansion.txt b/resources/dictionaries/ms/ambiguous_expansion.txt new file mode 100644 index 00000000..5695aa49 --- /dev/null +++ b/resources/dictionaries/ms/ambiguous_expansion.txt @@ -0,0 +1,6 @@ +mohd +dr +sek +keb +men +tudm diff --git a/resources/dictionaries/ms/building_types.txt b/resources/dictionaries/ms/building_types.txt new file mode 100644 index 00000000..2e9bb8c3 --- /dev/null +++ b/resources/dictionaries/ms/building_types.txt @@ -0,0 +1,8 @@ +apartment|apartments|aparment|apartmen|apt +building|bangunan|bgn +condominium|condominiums|kondominium|condo|kondo +flat|pangsapuri|rumah pangsa|p/puri +hall|dewan +house|rumah|rmh +residency|residensi|residen|residence|residences +tower|menara diff --git a/resources/dictionaries/ms/company_types.txt b/resources/dictionaries/ms/company_types.txt new file mode 100644 index 00000000..1161b99f --- /dev/null +++ b/resources/dictionaries/ms/company_types.txt @@ -0,0 +1,6 @@ +bank +company|syarikat +football club|kelab bola sepak +limited|berhad|bhd +private limited company|pte ltd|sendirian berhad|sdn bhd +society|persatuan diff --git a/resources/dictionaries/ms/directionals.txt b/resources/dictionaries/ms/directionals.txt index f704bd7f..4ee85545 100644 --- a/resources/dictionaries/ms/directionals.txt +++ b/resources/dictionaries/ms/directionals.txt @@ -1,9 +1,9 @@ -barat -barat daya -barat laut -selatan -tengah -tenggara -timur -timur laut -utara \ No newline at end of file +west|barat +southwest|barat daya +northwest|barat laut +south|selatan +center|central|centre|tengah +southeast|tenggara +east|timur +northeast|timur laut +north|utara diff --git a/resources/dictionaries/ms/given_names.txt b/resources/dictionaries/ms/given_names.txt new file mode 100644 index 00000000..c61193c1 --- /dev/null +++ b/resources/dictionaries/ms/given_names.txt @@ -0,0 +1 @@ +mohammad|muhammad|mohd diff --git a/resources/dictionaries/ms/level_types.txt b/resources/dictionaries/ms/level_types.txt new file mode 100644 index 00000000..b32d2cae --- /dev/null +++ b/resources/dictionaries/ms/level_types.txt @@ -0,0 +1 @@ +floor:level|aras|tingkat|tkt diff --git a/resources/dictionaries/ms/personal_titles.txt b/resources/dictionaries/ms/personal_titles.txt index 1032b091..342ff3da 100644 --- a/resources/dictionaries/ms/personal_titles.txt +++ b/resources/dictionaries/ms/personal_titles.txt @@ -1 +1,94 @@ -puteri \ No newline at end of file +yang di-pertuan agong|agong +yang di-pertuan besar +yang di-pertua negeri +tengku +syed +sharifah +raja +puteri|putri +putera|putra +sultan +tun +toh puan +tan sri +puan sri +datuk +datin +dato sri|dato seri +datuk seri +datin paduka seri +datin paduka seri panglima +datin paduka seri utama +datin paduka patinggi +dato +datin +to puan +datin paduka +dato wira +dato paduka +field marshal|fil marsyal +admiral of the fleet|laksamana armada +marshal of the air force|marsyal tentera udara +general|jeneral +admiral|laksamana +lieutenant general|leftenan jeneral +vice admiral|laksamana madya +major general|mejar jeneral +rear admiral|laksamana muda +brigadier general|brigedier jeneral +commodore|laksamana pertama +brigadier general|brigedier jeneral +colonel|kolonel +captain|kapten +lieutenant colonel|leftenan kolonel +commander|komander +major|mejar +lieutenant commander|leftenan komander +lieutenant|leftenan +sub-lieutenant|leftenan mad +second lieutenant|leftenan muda +junior sublieutenant|leftenan muda +sublieutenant|leftenan muda +midshipman|kadet kanan +officer cadet|pegawai kadet +warrant officer first class|pegawai waran kelas pertama +warrant officer second class|pegawai waran kelas kedua +staff sergeant|staff sarjan +chief petty officer|bintara kanan +flight sergeant|flait sarjan +sergeant|sarjan +petty officer|bintara muda +air sergeant|sarjan udara +corporal|korporal +leading rate|laskar kanan +air corporal|korporal udara +lance corporal|lans korporal +able seaman|laskar kelas pertama +leading aircraftman|laskar udara kanan +ordinary seaman|laskar kelas kedua +aircraftman first class|laskar udara kelas pertama +private|prebet +seaman|laskar muda +aircraftman second class|laskar udara kelas kedua +recruit|rekrut +seaman recruit|parajurit muda +ahli dewan undangan negeri|adun +chairman|pengerusi +deputy prime minister|timbalan perdana menteri +doctor|doktor|dr +judge|hakim +minister|menteri +mister|mr|encik|en +misses|mrs|puan|pn +miss|ms|cik +officer|pegawai +president|presiden +prime minister|perdana menteri +price|putera|putra +princess|puteri|putri +sir|tuan +representative|wakil +representatitves|wakil-wakil +vice chairman|vice chairperson|naib pengerusi +vice presiden|timbalan presiden +vice prime minister|timbalan perdana menteri diff --git a/resources/dictionaries/ms/place_names.txt b/resources/dictionaries/ms/place_names.txt index 3a8f8740..34e1e9a9 100644 --- a/resources/dictionaries/ms/place_names.txt +++ b/resources/dictionaries/ms/place_names.txt @@ -1,4 +1,139 @@ +academy|akademi +animal hospital|hospital haiwan +air force base|pangkalan tentera udara +airport|lapangan terbang|lpg terbang +aquarium|akuarium +arcade|arked +art gallery|galleri seni +arts center|arts centre|pusat seni +auditorium +bank +barrack|berek +beach|pantai +beauty salon|salon kecantikan +book store|kedai buku +cafe|kafe +campus|kampus +car park|tempat letak kereta|tapak letak kereta|parkir +casino|kasino +castle|istana +chiropractic|kiropraktik +church|gereja +cinema|panggung wayang gambar|pawagam +city hall|dewan bandar +clinic|klinik +club|kelab +clubhouse|rumah kelab +coffee|kopi +college|kolej +committee|jawatankuasa|jawatan kuasa +community center|pusat komuniti +community|komuniti +complex|kompleks +concert hall|dewan konsert +courthouse|mahkamah +culture center|pusat kebudayaan +dam|empangan +dance studio|studio tarian +development|pembangunan +distributor|pengedar +doctors|doktor-doktor +dormitory|asrama +elementary school|sekolah rendah|sekolah ren|sek rendah|sek ren +embassy|kedutaan +factory|kilang +farmer's market|pasar tani|pasar petani +farm|ladang pertanian|ladang +fire department|jabatan bomba +fire station|balai bomba +fitness center|pusat kecergasan +flats|rumah pangsa|pangsapuri|p/puri +fountain|mata air +garage|garajcesaasdf +garden|taman +gallery|galeri +gas station|stesen minyak +gate|pintu gerbang|pintu pagar +golf course|padang golf +golf club|kelab golfic +graveyard|tanah perkuburan +gym|gim|gynasium|gimnasium +hall|dewan +harbour|pelabuhan +headquarters|ibu pejabat|ibu pej +health center|pusat kesihatan +home|rumah kediaman +home for the aged|rumah orang tua +hospital|rumah sakit +hostel +house|rumah|rmh +ice cream|aiskrim +industrial park|taman perindustrian|taman industri|tmn ind|tmn perindustrian|tmn ind +institute|institut +jail|penjara +jetty|jeti +kindergarten|tadika +kitchen|dapur +league|liga +library|perpustakaan +marketplace|pasaran +mansion|rumah +medical|perubatan +mission|misi +monastery|biara +monument|monumen +municipal building|bangunan perbandaran +museum|muzium +music hall|dewan muzik +night club|kelab malam +nursing|kejururawatan +nursing center|pusat kejururawatan +nursing home|pusat penjagaan +middle school|junior high school|high school|sekolah menegah|sekolah men|sek menengah|sek men +laboratory|lab|makmal +market|pasar +movie theater|pawagam +national park|taman negara +nursery|nurseri +office|pejabat +office tower|menara pejabat +parking lot|tempat letak kereta +pediatric|pediatrik +performing arts center|pusat seni persembahan +pharmacy|farmasi +pier|dermaga +police department|jabatan polis +police station|balai polis +polytechnic|politeknik +port|pelabuhan +post office|pejabat pos +precinct|presint +pre school|pra sekolah +primary school|sekolah rendah|sek rendah|sekolah ren|sek ren +prison|penjara +pub +public pool|kolam renang awam +recycling|kitar semula +restaurant|restoran +room|bilik|blk +secondary school|sekolah menengah|sek menengah|sekolah men|sek men +shopping center|shopping mall|pusat membeli-belah +shop|kedai +social club|kelab sosial +store|stor +supermarket|pasar raya|pasaraya +swimming pool|kolam renang +tea|teh +terrace|teres +theater|panggung +tower|menara +town hall|dewan bandar +university|universiti +vault|bilik kebal +vet|doktor haiwan|dr haiwan +veterinary|veterinar +village|kampung|kampong|kg +youth center|pusat belia jelapang masjid -pasar -sekolah \ No newline at end of file +sekolah|sek diff --git a/resources/dictionaries/ms/post_office.txt b/resources/dictionaries/ms/post_office.txt new file mode 100644 index 00000000..d072caf5 --- /dev/null +++ b/resources/dictionaries/ms/post_office.txt @@ -0,0 +1,2 @@ +bag|beg +mail|mel|surat diff --git a/resources/dictionaries/ms/qualifiers.txt b/resources/dictionaries/ms/qualifiers.txt index 73c37104..39fda220 100644 --- a/resources/dictionaries/ms/qualifiers.txt +++ b/resources/dictionaries/ms/qualifiers.txt @@ -1,3 +1,6 @@ nusa penampang -pulau \ No newline at end of file +pulau +block|blok +section|seksyen +village|kampung|kampong|kg diff --git a/resources/dictionaries/ms/street_types.txt b/resources/dictionaries/ms/street_types.txt index 53d23a7a..bd52b9c7 100644 --- a/resources/dictionaries/ms/street_types.txt +++ b/resources/dictionaries/ms/street_types.txt @@ -1,17 +1,36 @@ awang|awg bulatan +changkat|ckt +cerunan|crn +dalaman|dlm +dataran|dtr dayang|dyg denai +gerbang|grb haji|hj -jalan|jln|jl +halaman|hlm +hilir|hlr +jalan|jln|jl|road|rd lapangan laluan -lebuhraya -lengkok -linkaran -lorong -kampong|kg +laman|lmn +langgak|lgk +lebuh|lbh|luh +lebuhraya|lbr +lengkok|lkk +linkaran|lkr +lintang|ltg +lorong|lrg +kampong|kampung|kg +medan|mdn pengiran|pg +persiaran|psn|psrn rapat simpang|spg -sngai|sg \ No newline at end of file +simpangan|spn +sisiran|ssr +solok|slk +sungai|sngai|sg +tingkat|tkt +taman|tmn +tepian|tpn diff --git a/resources/dictionaries/ms/synonyms.txt b/resources/dictionaries/ms/synonyms.txt new file mode 100644 index 00000000..1d3c0bc3 --- /dev/null +++ b/resources/dictionaries/ms/synonyms.txt @@ -0,0 +1,25 @@ +timur|timor +aras|tingkat|tkt +air|ayer +kampung|kampong|kg +itam|hitam +tanjung|tanjong +sri|seri +tasik|tasek +dwitasik|dwitasek +fasa|phasa +selasih|selaseh +putera|putra +puteri|putri +kecil|kechil +sungai|sungei|sg +teluk|telok +yusuf|yusof +batu|bt +bukit|bkt +buluh|buloh +jelutung|jelutong +tanduk|tandok +chempaka|cempaka +suleiman|sulaiman +mohammad|muhammad From 20b87ba5c8c3e220bd090219e103ffc3a2110dd3 Mon Sep 17 00:00:00 2001 From: jeffrey04 Date: Fri, 30 Sep 2016 10:01:13 +0800 Subject: [PATCH 02/12] removing ambiguous_expansion(s).txt --- resources/dictionaries/ms/ambiguous_expansion.txt | 6 ------ 1 file changed, 6 deletions(-) delete mode 100644 resources/dictionaries/ms/ambiguous_expansion.txt diff --git a/resources/dictionaries/ms/ambiguous_expansion.txt b/resources/dictionaries/ms/ambiguous_expansion.txt deleted file mode 100644 index 5695aa49..00000000 --- a/resources/dictionaries/ms/ambiguous_expansion.txt +++ /dev/null @@ -1,6 +0,0 @@ -mohd -dr -sek -keb -men -tudm From f43ba7fe6349cf5abc2c40da50d4bd248a45ac84 Mon Sep 17 00:00:00 2001 From: jeffrey04 Date: Fri, 30 Sep 2016 10:14:25 +0800 Subject: [PATCH 03/12] removing english words from dictionary --- resources/dictionaries/ms/building_types.txt | 14 +- resources/dictionaries/ms/company_types.txt | 10 +- resources/dictionaries/ms/directionals.txt | 18 +- resources/dictionaries/ms/level_types.txt | 2 +- resources/dictionaries/ms/personal_titles.txt | 127 +++++---- resources/dictionaries/ms/place_names.txt | 262 +++++++++--------- resources/dictionaries/ms/post_office.txt | 4 +- resources/dictionaries/ms/qualifiers.txt | 6 +- 8 files changed, 220 insertions(+), 223 deletions(-) diff --git a/resources/dictionaries/ms/building_types.txt b/resources/dictionaries/ms/building_types.txt index 2e9bb8c3..a23cc082 100644 --- a/resources/dictionaries/ms/building_types.txt +++ b/resources/dictionaries/ms/building_types.txt @@ -1,8 +1,8 @@ -apartment|apartments|aparment|apartmen|apt -building|bangunan|bgn -condominium|condominiums|kondominium|condo|kondo +apartment|aparment|apartmen|apt +bangunan|bgn +kondominium|kondo flat|pangsapuri|rumah pangsa|p/puri -hall|dewan -house|rumah|rmh -residency|residensi|residen|residence|residences -tower|menara +dewan +rumah|rmh +residensi|residen +menara diff --git a/resources/dictionaries/ms/company_types.txt b/resources/dictionaries/ms/company_types.txt index 1161b99f..81762cd8 100644 --- a/resources/dictionaries/ms/company_types.txt +++ b/resources/dictionaries/ms/company_types.txt @@ -1,6 +1,6 @@ bank -company|syarikat -football club|kelab bola sepak -limited|berhad|bhd -private limited company|pte ltd|sendirian berhad|sdn bhd -society|persatuan +syarikat +kelab bola sepak +berhad|bhd +sendirian berhad|sdn bhd +persatuan diff --git a/resources/dictionaries/ms/directionals.txt b/resources/dictionaries/ms/directionals.txt index 4ee85545..d6da4091 100644 --- a/resources/dictionaries/ms/directionals.txt +++ b/resources/dictionaries/ms/directionals.txt @@ -1,9 +1,9 @@ -west|barat -southwest|barat daya -northwest|barat laut -south|selatan -center|central|centre|tengah -southeast|tenggara -east|timur -northeast|timur laut -north|utara +barat +barat daya +barat laut +selatan +tengah +tenggara +timur +timur laut +utara diff --git a/resources/dictionaries/ms/level_types.txt b/resources/dictionaries/ms/level_types.txt index b32d2cae..4296e606 100644 --- a/resources/dictionaries/ms/level_types.txt +++ b/resources/dictionaries/ms/level_types.txt @@ -1 +1 @@ -floor:level|aras|tingkat|tkt +aras|tingkat|tkt diff --git a/resources/dictionaries/ms/personal_titles.txt b/resources/dictionaries/ms/personal_titles.txt index 342ff3da..75e14255 100644 --- a/resources/dictionaries/ms/personal_titles.txt +++ b/resources/dictionaries/ms/personal_titles.txt @@ -26,69 +26,66 @@ to puan datin paduka dato wira dato paduka -field marshal|fil marsyal -admiral of the fleet|laksamana armada -marshal of the air force|marsyal tentera udara -general|jeneral -admiral|laksamana -lieutenant general|leftenan jeneral -vice admiral|laksamana madya -major general|mejar jeneral -rear admiral|laksamana muda -brigadier general|brigedier jeneral -commodore|laksamana pertama -brigadier general|brigedier jeneral -colonel|kolonel -captain|kapten -lieutenant colonel|leftenan kolonel -commander|komander -major|mejar -lieutenant commander|leftenan komander -lieutenant|leftenan -sub-lieutenant|leftenan mad -second lieutenant|leftenan muda -junior sublieutenant|leftenan muda -sublieutenant|leftenan muda -midshipman|kadet kanan -officer cadet|pegawai kadet -warrant officer first class|pegawai waran kelas pertama -warrant officer second class|pegawai waran kelas kedua -staff sergeant|staff sarjan -chief petty officer|bintara kanan -flight sergeant|flait sarjan -sergeant|sarjan -petty officer|bintara muda -air sergeant|sarjan udara -corporal|korporal -leading rate|laskar kanan -air corporal|korporal udara -lance corporal|lans korporal -able seaman|laskar kelas pertama -leading aircraftman|laskar udara kanan -ordinary seaman|laskar kelas kedua -aircraftman first class|laskar udara kelas pertama -private|prebet -seaman|laskar muda -aircraftman second class|laskar udara kelas kedua -recruit|rekrut -seaman recruit|parajurit muda +fil marsyal +laksamana armada +marsyal tentera udara +jeneral +laksamana +leftenan jeneral +laksamana madya +mejar jeneral +laksamana muda +brigedier jeneral +laksamana pertama +kolonel +kapten +leftenan kolonel +komander +mejar +leftenan komander +leftenan +leftenan madya +leftenan muda +kadet kanan +pegawai kadet +pegawai waran kelas pertama +pegawai waran kelas kedua +staff sarjan +bintara kanan +flait sarjan +sarjan +bintara muda +sarjan udara +korporal +laskar kanan +korporal udara +lans korporal +laskar kelas pertama +laskar udara kanan +laskar kelas kedua +laskar udara kelas pertama +prebet +laskar muda +laskar udara kelas kedua +rekrut +parajurit muda ahli dewan undangan negeri|adun -chairman|pengerusi -deputy prime minister|timbalan perdana menteri -doctor|doktor|dr -judge|hakim -minister|menteri -mister|mr|encik|en -misses|mrs|puan|pn -miss|ms|cik -officer|pegawai -president|presiden -prime minister|perdana menteri -price|putera|putra -princess|puteri|putri -sir|tuan -representative|wakil -representatitves|wakil-wakil -vice chairman|vice chairperson|naib pengerusi -vice presiden|timbalan presiden -vice prime minister|timbalan perdana menteri +pengerusi +timbalan perdana menteri +doktor|dr +hakim +menteri +mr|encik|en +mrs|puan|pn +ms|cik +pegawai +presiden +perdana menteri +putera|putra +puteri|putri +tuan +wakil +wakil-wakil +naib pengerusi +timbalan presiden +timbalan perdana menteri diff --git a/resources/dictionaries/ms/place_names.txt b/resources/dictionaries/ms/place_names.txt index 34e1e9a9..4a866bec 100644 --- a/resources/dictionaries/ms/place_names.txt +++ b/resources/dictionaries/ms/place_names.txt @@ -1,139 +1,139 @@ -academy|akademi -animal hospital|hospital haiwan -air force base|pangkalan tentera udara -airport|lapangan terbang|lpg terbang -aquarium|akuarium -arcade|arked -art gallery|galleri seni -arts center|arts centre|pusat seni +akademi +hospital haiwan +pangkalan tentera udara +lapangan terbang|lpg terbang +akuarium +arked +galleri seni +pusat seni auditorium bank -barrack|berek -beach|pantai -beauty salon|salon kecantikan -book store|kedai buku -cafe|kafe -campus|kampus -car park|tempat letak kereta|tapak letak kereta|parkir -casino|kasino -castle|istana -chiropractic|kiropraktik -church|gereja -cinema|panggung wayang gambar|pawagam -city hall|dewan bandar -clinic|klinik -club|kelab -clubhouse|rumah kelab -coffee|kopi -college|kolej -committee|jawatankuasa|jawatan kuasa -community center|pusat komuniti -community|komuniti -complex|kompleks -concert hall|dewan konsert -courthouse|mahkamah -culture center|pusat kebudayaan -dam|empangan -dance studio|studio tarian -development|pembangunan -distributor|pengedar -doctors|doktor-doktor -dormitory|asrama -elementary school|sekolah rendah|sekolah ren|sek rendah|sek ren -embassy|kedutaan -factory|kilang -farmer's market|pasar tani|pasar petani -farm|ladang pertanian|ladang -fire department|jabatan bomba -fire station|balai bomba -fitness center|pusat kecergasan -flats|rumah pangsa|pangsapuri|p/puri -fountain|mata air -garage|garajcesaasdf -garden|taman -gallery|galeri -gas station|stesen minyak -gate|pintu gerbang|pintu pagar -golf course|padang golf -golf club|kelab golfic -graveyard|tanah perkuburan -gym|gim|gynasium|gimnasium -hall|dewan -harbour|pelabuhan -headquarters|ibu pejabat|ibu pej -health center|pusat kesihatan -home|rumah kediaman -home for the aged|rumah orang tua +berek +pantai +salon kecantikan +kedai buku +kafe +kampus +tempat letak kereta|tapak letak kereta|parkir +kasino +istana +kiropraktik +gereja +panggung wayang gambar|pawagam +dewan bandar +klinik +kelab +rumah kelab +kopi +kolej +jawatankuasa|jawatan kuasa +pusat komuniti +komuniti +kompleks +dewan konsert +mahkamah +pusat kebudayaan +empangan +studio tarian +pembangunan +pengedar +doktor-doktor +asrama +sekolah rendah|sekolah ren|sek rendah|sek ren +kedutaan +kilang +pasar tani|pasar petani +ladang pertanian|ladang +jabatan bomba +balai bomba +pusat kecergasan +flat|rumah pangsa|pangsapuri|p/puri +mata air +garaj +taman +galeri +stesen minyak +pintu gerbang|pintu pagar +padang golf +kelab golf +tanah perkuburan +gim|gimnasium +dewan +pelabuhan +ibu pejabat|ibu pej +pusat kesihatan +rumah kediaman +rumah orang tua hospital|rumah sakit hostel -house|rumah|rmh -ice cream|aiskrim -industrial park|taman perindustrian|taman industri|tmn ind|tmn perindustrian|tmn ind -institute|institut -jail|penjara -jetty|jeti -kindergarten|tadika -kitchen|dapur -league|liga -library|perpustakaan -marketplace|pasaran -mansion|rumah -medical|perubatan -mission|misi -monastery|biara -monument|monumen -municipal building|bangunan perbandaran -museum|muzium -music hall|dewan muzik -night club|kelab malam -nursing|kejururawatan -nursing center|pusat kejururawatan -nursing home|pusat penjagaan -middle school|junior high school|high school|sekolah menegah|sekolah men|sek menengah|sek men -laboratory|lab|makmal -market|pasar -movie theater|pawagam -national park|taman negara -nursery|nurseri -office|pejabat -office tower|menara pejabat -parking lot|tempat letak kereta -pediatric|pediatrik -performing arts center|pusat seni persembahan -pharmacy|farmasi -pier|dermaga -police department|jabatan polis -police station|balai polis -polytechnic|politeknik -port|pelabuhan -post office|pejabat pos -precinct|presint -pre school|pra sekolah -primary school|sekolah rendah|sek rendah|sekolah ren|sek ren -prison|penjara +rumah|rmh +aiskrim +taman perindustrian|taman industri|tmn ind|tmn perindustrian|tmn ind +institut +penjara +jeti +tadika +dapur +liga +perpustakaan +pasaran +rumah +perubatan +misi +biara +monumen +bangunan perbandaran +muzium +dewan muzik +kelab malam +kejururawatan +pusat kejururawatan +pusat penjagaan +junior high school|high school|sekolah menegah|sekolah men|sek menengah|sek men +makmal +pasar +pawagam +taman negara +nurseri +pejabat +menara pejabat +tempat letak kereta +pediatrik +pusat seni persembahan +farmasi +dermaga +jabatan polis +balai polis +politeknik +pelabuhan +pejabat pos +presint +pra sekolah +sekolah rendah|sek rendah|sekolah ren|sek ren +penjara pub -public pool|kolam renang awam -recycling|kitar semula -restaurant|restoran -room|bilik|blk -secondary school|sekolah menengah|sek menengah|sekolah men|sek men -shopping center|shopping mall|pusat membeli-belah -shop|kedai -social club|kelab sosial -store|stor -supermarket|pasar raya|pasaraya -swimming pool|kolam renang -tea|teh -terrace|teres -theater|panggung -tower|menara -town hall|dewan bandar -university|universiti -vault|bilik kebal -vet|doktor haiwan|dr haiwan -veterinary|veterinar -village|kampung|kampong|kg -youth center|pusat belia +kolam renang awam +kitar semula +restoran +bilik|blk +sekolah menengah|sek menengah|sekolah men|sek men +shopping mall|pusat membeli-belah +kedai +kelab sosial +stor +pasar raya|pasaraya +kolam renang +teh +teres +panggung +menara +dewan bandar +universiti +bilik kebal +doktor haiwan|dr haiwan +veterinar +kampung|kampong|kg +pusat belia jelapang masjid sekolah|sek diff --git a/resources/dictionaries/ms/post_office.txt b/resources/dictionaries/ms/post_office.txt index d072caf5..99bd95bf 100644 --- a/resources/dictionaries/ms/post_office.txt +++ b/resources/dictionaries/ms/post_office.txt @@ -1,2 +1,2 @@ -bag|beg -mail|mel|surat +beg +mel|surat diff --git a/resources/dictionaries/ms/qualifiers.txt b/resources/dictionaries/ms/qualifiers.txt index 39fda220..53eedc24 100644 --- a/resources/dictionaries/ms/qualifiers.txt +++ b/resources/dictionaries/ms/qualifiers.txt @@ -1,6 +1,6 @@ nusa penampang pulau -block|blok -section|seksyen -village|kampung|kampong|kg +blok +seksyen +kampung|kampong|kg From 8ae8340beeea135c775dcbe6c910be8c20f9f89c Mon Sep 17 00:00:00 2001 From: jeffrey04 Date: Fri, 30 Sep 2016 10:18:04 +0800 Subject: [PATCH 04/12] remove shopping mall from list --- resources/dictionaries/ms/place_names.txt | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/resources/dictionaries/ms/place_names.txt b/resources/dictionaries/ms/place_names.txt index 4a866bec..87c0986d 100644 --- a/resources/dictionaries/ms/place_names.txt +++ b/resources/dictionaries/ms/place_names.txt @@ -117,7 +117,7 @@ kitar semula restoran bilik|blk sekolah menengah|sek menengah|sekolah men|sek men -shopping mall|pusat membeli-belah +pusat membeli-belah kedai kelab sosial stor From f5477a73699d703d1c3f6bb1d47bc2dab748dc68 Mon Sep 17 00:00:00 2001 From: jeffrey04 Date: Tue, 4 Oct 2016 11:29:28 +0800 Subject: [PATCH 05/12] each term should be in a separate line --- resources/dictionaries/ms/building_types.txt | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/resources/dictionaries/ms/building_types.txt b/resources/dictionaries/ms/building_types.txt index a23cc082..d702e524 100644 --- a/resources/dictionaries/ms/building_types.txt +++ b/resources/dictionaries/ms/building_types.txt @@ -1,7 +1,7 @@ apartment|aparment|apartmen|apt bangunan|bgn kondominium|kondo -flat|pangsapuri|rumah pangsa|p/puri +pangsapuri|rumah pangsa|p/puri dewan rumah|rmh residensi|residen From 57210bd6570c4291e17a64957b04866ead68628d Mon Sep 17 00:00:00 2001 From: jeffrey04 Date: Tue, 4 Oct 2016 11:30:09 +0800 Subject: [PATCH 06/12] each term should be in separate lines --- resources/dictionaries/ms/level_types.txt | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/resources/dictionaries/ms/level_types.txt b/resources/dictionaries/ms/level_types.txt index 4296e606..30a4bf18 100644 --- a/resources/dictionaries/ms/level_types.txt +++ b/resources/dictionaries/ms/level_types.txt @@ -1 +1,2 @@ -aras|tingkat|tkt +aras +tingkat|tkt From b2305b574d15cdb18870593782e50f5a0c36304d Mon Sep 17 00:00:00 2001 From: jeffrey04 Date: Tue, 4 Oct 2016 11:30:28 +0800 Subject: [PATCH 07/12] removing english abbr --- resources/dictionaries/ms/personal_titles.txt | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/resources/dictionaries/ms/personal_titles.txt b/resources/dictionaries/ms/personal_titles.txt index 75e14255..4b7ac4cf 100644 --- a/resources/dictionaries/ms/personal_titles.txt +++ b/resources/dictionaries/ms/personal_titles.txt @@ -75,9 +75,9 @@ timbalan perdana menteri doktor|dr hakim menteri -mr|encik|en -mrs|puan|pn -ms|cik +encik|en +puan|pn +cik pegawai presiden perdana menteri From 244dbbdd4ac590facb97f49535d12d61d3a37f24 Mon Sep 17 00:00:00 2001 From: Al Date: Thu, 6 Oct 2016 20:27:15 -0400 Subject: [PATCH 08/12] [fix] separating synonyms that are for different words --- resources/dictionaries/ms/building_types.txt | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/resources/dictionaries/ms/building_types.txt b/resources/dictionaries/ms/building_types.txt index d702e524..8d167f99 100644 --- a/resources/dictionaries/ms/building_types.txt +++ b/resources/dictionaries/ms/building_types.txt @@ -1,8 +1,9 @@ apartment|aparment|apartmen|apt bangunan|bgn kondominium|kondo -pangsapuri|rumah pangsa|p/puri +pangsapuri|p/puri dewan rumah|rmh +rumah pangsa residensi|residen menara From 2c48acd680a0e702dd947bda7e8bb435e229fdf7 Mon Sep 17 00:00:00 2001 From: Al Date: Thu, 6 Oct 2016 20:28:44 -0400 Subject: [PATCH 09/12] [dictionaries] removing flat/rumah pangsa/pangsapuri from place_names, aliasing gim to gimnasium rather than the other way around, removing duplicate/mixed English + Malay line --- resources/dictionaries/ms/place_names.txt | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/resources/dictionaries/ms/place_names.txt b/resources/dictionaries/ms/place_names.txt index 87c0986d..5ab138d9 100644 --- a/resources/dictionaries/ms/place_names.txt +++ b/resources/dictionaries/ms/place_names.txt @@ -47,7 +47,6 @@ ladang pertanian|ladang jabatan bomba balai bomba pusat kecergasan -flat|rumah pangsa|pangsapuri|p/puri mata air garaj taman @@ -57,7 +56,7 @@ pintu gerbang|pintu pagar padang golf kelab golf tanah perkuburan -gim|gimnasium +gimnasium|gim dewan pelabuhan ibu pejabat|ibu pej @@ -89,7 +88,6 @@ kelab malam kejururawatan pusat kejururawatan pusat penjagaan -junior high school|high school|sekolah menegah|sekolah men|sek menengah|sek men makmal pasar pawagam From c4e147ed20fdc1b7c729951a11c67a96e54895ad Mon Sep 17 00:00:00 2001 From: Al Date: Thu, 6 Oct 2016 20:29:09 -0400 Subject: [PATCH 10/12] [fix] separating words that have different roots --- resources/dictionaries/ms/post_office.txt | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/resources/dictionaries/ms/post_office.txt b/resources/dictionaries/ms/post_office.txt index 99bd95bf..c28a853d 100644 --- a/resources/dictionaries/ms/post_office.txt +++ b/resources/dictionaries/ms/post_office.txt @@ -1,2 +1,3 @@ beg -mel|surat +mel +surat From 5f42e66f311fe2e44639c1ec909bfac4f3861785 Mon Sep 17 00:00:00 2001 From: Al Date: Thu, 6 Oct 2016 20:29:35 -0400 Subject: [PATCH 11/12] [fix] removing road/rd from the synonyms list for jalan as they're covered by the English dictionaries --- resources/dictionaries/ms/street_types.txt | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/resources/dictionaries/ms/street_types.txt b/resources/dictionaries/ms/street_types.txt index bd52b9c7..387dc2f4 100644 --- a/resources/dictionaries/ms/street_types.txt +++ b/resources/dictionaries/ms/street_types.txt @@ -10,7 +10,7 @@ gerbang|grb haji|hj halaman|hlm hilir|hlr -jalan|jln|jl|road|rd +jalan|jln|jl lapangan laluan laman|lmn From 03d0afb8209a606fb86ceb5d4abaa3726e940e16 Mon Sep 17 00:00:00 2001 From: Al Date: Thu, 6 Oct 2016 20:30:48 -0400 Subject: [PATCH 12/12] [fix] removing level types and given names from synonyms since they're already covered --- resources/dictionaries/ms/synonyms.txt | 2 -- 1 file changed, 2 deletions(-) diff --git a/resources/dictionaries/ms/synonyms.txt b/resources/dictionaries/ms/synonyms.txt index 1d3c0bc3..9d1129e4 100644 --- a/resources/dictionaries/ms/synonyms.txt +++ b/resources/dictionaries/ms/synonyms.txt @@ -1,5 +1,4 @@ timur|timor -aras|tingkat|tkt air|ayer kampung|kampong|kg itam|hitam @@ -22,4 +21,3 @@ jelutung|jelutong tanduk|tandok chempaka|cempaka suleiman|sulaiman -mohammad|muhammad