X-Git-Url: https://code.communitydata.science/covid19.git/blobdiff_plain/09d171608f699eafe6528325938a892937a65302..98b07b8098611287eaa775b09622d1f3514303c8:/keywords/src/compile_transliterated_phrases.sh

diff --git a/keywords/src/compile_transliterated_phrases.sh b/keywords/src/compile_transliterated_phrases.sh
new file mode 100755
index 0000000..6d633ee
--- /dev/null
+++ b/keywords/src/compile_transliterated_phrases.sh
@@ -0,0 +1,16 @@
+#!/bin/bash
+
+# For now these scripts don't accept command line arguments. It's an MVP
+
+echo "Reading Google trends"
+python3 collect_trends.py
+
+echo "Searching for Wikidata entities using base_terms.txt"
+python3 wikidata_search.py ../resources/base_terms.txt --output ../output/intermediate/wikidata_search_results.csv
+
+echo "Searching for Wikidata entities using Google trends"
+python3 wikidata_search.py ../output/intermediate/related_searches_rising.csv ../output/intermediate/related_searches_top.csv --use-gtrends --output ../output/intermediate/wikidata_search_results_from_gtrends.csv
+
+echo "Finding transliterations from Wikidata using sparql"
+python3 wikidata_transliterations.py  ../output/intermediate/wikidata_search_results_from_gtrends.csv  ../output/intermediate/wikidata_search_results.csv --topN 10 20 --output ../output/csv/$(date '+%Y-%m-%d')_wikidata_entity_labels.csv
+