The Visual Global Knowledge Graph (VGKG) has processed more than half a billion worldwide online news images since December 2015, cataloging the visual narratives of the global media landscape. What are the most common labels that describe the objects and activities of the global news landscape? The query below compiles the top tags of the last 12 months, spanning from July 2019 through present:
SELECT LOWER(REGEXP_REPLACE(Rec, r'<FIELD>.*', '')) Label, count(1) Count FROM `gdelt-bq.gdeltv2.cloudvision_partitioned`, UNNEST(SPLIT(Labels, '<RECORD>')) Rec where DATE(_PARTITIONTIME) >= "2019-07-01" group by Label order by Count desc
The final query took just 10 seconds to process 34GB of data. You can download the final resulting spreadsheet.