added initial version of revision-scraper

[covid19.git] / transliterations / src / compile_transliterated_phrases.sh

diff --git a/transliterations/src/compile_transliterated_phrases.sh b/transliterations/src/compile_transliterated_phrases.sh
index 8af82f7bff7a837e909bae40341e8085d624def5..6d633eebbbcbcfa8f9b523aa8f016ddd3f114da7 100755 (executable)
--- a/transliterations/src/compile_transliterated_phrases.sh
+++ b/transliterations/src/compile_transliterated_phrases.sh
@@ -2,5 +2,15 @@
 
 # For now these scripts don't accept command line arguments. It's an MVP
 
 
 # For now these scripts don't accept command line arguments. It's an MVP
 

-python3 wikidata_search.py
-python3 wikidata_transliterations.py
+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
+