[dictionaries] adding common Hindi tokens
Reviewed the Hindi ngrams list, identified a number of common items which could be added to the model
This commit is contained in:
5
resources/dictionaries/hi/building_types.txt
Normal file
5
resources/dictionaries/hi/building_types.txt
Normal file
@@ -0,0 +1,5 @@
|
|||||||
|
mandir|मन्दिर|मंदिर
|
||||||
|
station
|
||||||
|
police station
|
||||||
|
post office
|
||||||
|
office
|
||||||
4
resources/dictionaries/hi/directionals.txt
Normal file
4
resources/dictionaries/hi/directionals.txt
Normal file
@@ -0,0 +1,4 @@
|
|||||||
|
dakshin|दक्षिण
|
||||||
|
uttar|उत्तर
|
||||||
|
poorva|poorav|पूर्व
|
||||||
|
paschim|पश्चिम
|
||||||
1
resources/dictionaries/hi/people.txt
Normal file
1
resources/dictionaries/hi/people.txt
Normal file
@@ -0,0 +1 @@
|
|||||||
|
mahatma gandhi|mg|m g
|
||||||
24
resources/dictionaries/hi/personal_titles.txt
Normal file
24
resources/dictionaries/hi/personal_titles.txt
Normal file
@@ -0,0 +1,24 @@
|
|||||||
|
baba
|
||||||
|
babu
|
||||||
|
bhagat
|
||||||
|
guru
|
||||||
|
jagirdar
|
||||||
|
maharaja|maharaj
|
||||||
|
mahatma|महात्मा
|
||||||
|
pandit
|
||||||
|
raja
|
||||||
|
rajarshi
|
||||||
|
rajkumar
|
||||||
|
rajkumari
|
||||||
|
rani
|
||||||
|
rishi
|
||||||
|
sahib
|
||||||
|
sant
|
||||||
|
sardar
|
||||||
|
senapati
|
||||||
|
shah
|
||||||
|
shrimati|smt|srimathi|श्रीमती
|
||||||
|
shri|shree|sri|श्री
|
||||||
|
sushri
|
||||||
|
swami
|
||||||
|
ustad
|
||||||
3
resources/dictionaries/hi/qualifiers.txt
Normal file
3
resources/dictionaries/hi/qualifiers.txt
Normal file
@@ -0,0 +1,3 @@
|
|||||||
|
nagar|नगर
|
||||||
|
colony
|
||||||
|
cantonment|cantt
|
||||||
@@ -1,3 +1,4 @@
|
|||||||
bazaar|bazar
|
bazaar|bazar
|
||||||
marg
|
marg
|
||||||
nagar
|
flyover
|
||||||
|
रोड
|
||||||
Reference in New Issue
Block a user