X-Git-Url: https://code.communitydata.science/covid19.git/blobdiff_plain/6905a5204416d2d7e2c17e9dfb0fd77852783ab2..72bf7bcd3787ffbda4ec2c47204896483e8069c9:/transliterations/src/compile_transliterated_phrases.sh diff --git a/transliterations/src/compile_transliterated_phrases.sh b/transliterations/src/compile_transliterated_phrases.sh index 8af82f7..6d633ee 100755 --- 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 -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 +