We are incredibly excited to announce today the release of Version 2.0 of the GDELT Visual Global Knowledge Graph (VGKG), powered by Google's Cloud Vision API! With the release of VGKG 2.0, the original 1.0 data streams will continue to be available and your existing workflows will continue to work without interruption.
With the release of 2.0, you will see a new "EXIF" field in the RawJSON field of each record that contains a wealth of metadata about the image containing every available embedded field in EXIF, IPTC, XMP and all other major metadata formats. Note that not all images may have embedded metadata and that these fields are extracted as-is, meaning you can encounter strange and non-standard results in some images. Non-ASCII characters are Unicode escaped.
In addition, a new "ImageProperties" field in the RawJSON adds several new fields, including the timestamp of when GDELT first encountered the image, the URL of the article it was first seen in, estimated languages of any text that might be found in the image, its MD5 hash and a set of perceptual hashes for visual similarity comparisons.
Finally, as the first GDELT 3.0 release, the VGKG 2.0 data stream updates every 60 seconds, allowing you near-realtime access to a codified view of global visual narratives. Just fetch http://data.gdeltproject.org/gdeltv3/vgkg/lastupdate.txt every minute to get the latest update file. If this stream is too fast for you, you can simply download the original VGKG 1.0 stream, which updates every 15 minutes – it is simply a rollup of these 1 minute files over the last 15 minutes.
We're immensely excited about this new release and cannot wait to see what you're able to accomplish with it!