From 18b298668b54eaba0bce75e4be8501751e7efe4a Mon Sep 17 00:00:00 2001
From: karanj <karanj@gmail.com>
Date: Thu, 16 Feb 2023 21:45:06 +1100
Subject: [PATCH] [dictionaries] adding common Hindi tokens

Reviewed the Hindi ngrams list, identified a number of common items which could be added to the model
---
 resources/dictionaries/hi/building_types.txt  |  5 ++++
 resources/dictionaries/hi/directionals.txt    |  4 ++++
 resources/dictionaries/hi/people.txt          |  1 +
 resources/dictionaries/hi/personal_titles.txt | 24 +++++++++++++++++++
 resources/dictionaries/hi/qualifiers.txt      |  3 +++
 resources/dictionaries/hi/street_types.txt    |  3 ++-
 6 files changed, 39 insertions(+), 1 deletion(-)
 create mode 100644 resources/dictionaries/hi/building_types.txt
 create mode 100644 resources/dictionaries/hi/directionals.txt
 create mode 100644 resources/dictionaries/hi/people.txt
 create mode 100644 resources/dictionaries/hi/personal_titles.txt
 create mode 100644 resources/dictionaries/hi/qualifiers.txt

diff --git a/resources/dictionaries/hi/building_types.txt b/resources/dictionaries/hi/building_types.txt
new file mode 100644
index 000000000..180e36062
--- /dev/null
+++ b/resources/dictionaries/hi/building_types.txt
@@ -0,0 +1,5 @@
+mandir|मन्दिर|मंदिर
+station
+police station
+post office
+office
\ No newline at end of file
diff --git a/resources/dictionaries/hi/directionals.txt b/resources/dictionaries/hi/directionals.txt
new file mode 100644
index 000000000..4dfff2006
--- /dev/null
+++ b/resources/dictionaries/hi/directionals.txt
@@ -0,0 +1,4 @@
+dakshin|दक्षिण
+uttar|उत्तर
+poorva|poorav|पूर्व
+paschim|पश्चिम
\ No newline at end of file
diff --git a/resources/dictionaries/hi/people.txt b/resources/dictionaries/hi/people.txt
new file mode 100644
index 000000000..b79d681f5
--- /dev/null
+++ b/resources/dictionaries/hi/people.txt
@@ -0,0 +1 @@
+mahatma gandhi|mg|m g
\ No newline at end of file
diff --git a/resources/dictionaries/hi/personal_titles.txt b/resources/dictionaries/hi/personal_titles.txt
new file mode 100644
index 000000000..6f60d0cda
--- /dev/null
+++ b/resources/dictionaries/hi/personal_titles.txt
@@ -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
\ No newline at end of file
diff --git a/resources/dictionaries/hi/qualifiers.txt b/resources/dictionaries/hi/qualifiers.txt
new file mode 100644
index 000000000..ba5b38f7f
--- /dev/null
+++ b/resources/dictionaries/hi/qualifiers.txt
@@ -0,0 +1,3 @@
+nagar|नगर
+colony
+cantonment|cantt
\ No newline at end of file
diff --git a/resources/dictionaries/hi/street_types.txt b/resources/dictionaries/hi/street_types.txt
index fde3a96b2..b5cb55286 100644
--- a/resources/dictionaries/hi/street_types.txt
+++ b/resources/dictionaries/hi/street_types.txt
@@ -1,3 +1,4 @@
 bazaar|bazar
 marg
-nagar
\ No newline at end of file
+flyover
+रोड
\ No newline at end of file