One of the most amazing things about Google's BigQuery service is its nearly infinite scalability and the sheer speed at which it can perform tasks even on larger datasets. In just the last few weeks we've used BigQuery to explore billions of location mentions to map the hidden geography of language, mapped 6.2 billion location references […]
Vox' Albert Chang published another fantastic data-driven analysis of the parallel media universes using the Television Explorer. Read The Full Article.
One of the most powerful features of GDELT Summary is the ability it gives you to search across the world's languages, offering familiar keyword searching across the English machine translations of worldwide news coverage in 65 languages. This ensures you are able to find coverage regardless of what country it was published in or what language […]
When displaying human-friendly results based on GDELT's data, its often useful to show the formal name of a news outlet instead of its domain name, such as "CNN" instead of "cnn.com" or "Washington Post" instead of "washingtonpost.com". Towards that end, we've released a new dataset that attempts to identify the short formal name of each […]
It was just this past January that the DOC 2.0 API was extended from searching just the last three months of coverage to searching a rolling window of the past full year of coverage. We're excited to announce today that instead of a rolling window of one year of coverage, the DOC 2.0 API will now […]
Combining fulltext geocoding with the power of the cloud we can transform 126 billion words of news coverage from a quarter billion articles into 1.5 billion coordinates and with a single line of code map the hidden geographic affinities of language that form the lens through which we see the world. Read The Full Article.
Having spent more than a decade mapping the world through the lens of textual geography, we've become fascinated with the hidden geography of language, especially the natural geographic affinity of individual words. Take the word "love." What might it look like if one was to take a large sample of all of the worldwide news […]
Even as Facebook takes steps to reign in mass academic harvesting of private user data, those efforts are running up against the reality of years of freewheeling harvesting and data sharing and the company's historical reluctance to constrain academic misuse of its two billion users' data. Read The Full Article.
When mapping the locations of the world’s news outlets, the difference between geographic focus and geographic location offers a powerful lesson in how we see the world and the different lenses different datasets offer us. Read The Full Article.
Social media has a terrorism problem, but the platforms’ overreliance on AI is blinding them, public confusion and lack of detail means we no idea whether they are actually working and the lack of accountability means the democratic vision of the early web is giving way to dystopian nightmare. Read The Full Article.
GDELT is used as one of the datasets being analyzed in a paper titled "Online Continuous-Time Tensor Factorization Based on Pairwise Interactive Point Processes" by Hongteng Xu, Dixin Luo and Lawrence Carin of Infinia ML, Inc and Duke University's Department of Electrical and Computer Engineering will be presented at the 2018 International Joint Conference on Artificial […]
Given GDELT's incredibly unique ability to peer across the world's languages courtesy of its mass machine translation infrastructure, a common application is to compare coverage of major events in different countries. Doing so requires the ability to place news outlets geographically in space. At first glance, placing a news outlet on a map might seem […]
In 2015 we created a map of the geography of the world's media as seen through a year of GDELT's eyes, yielding a clickable map where you could see the top 5 outlets publishing the greatest volume of content about each city on earth. Today we've updated that analysis, enhancing it from a single absolute […]
While we tend to think of geographic locations as having just one name, in reality they can have dozens or even hundreds of alternative names, name variants and name transliterations. When geocoding a news article, GDELT records the location name as it appeared in the article, which is why we recommend using the FeatureID field […]
Using BigQuery To Make A 38 Year Timeline Of World Conflict From Half A Billion Records In 3 Seconds
Three years ago we showed off the results of a singularly incredible BigQuery query, courtesy of the one and only Felipe Hoffa, that processes the entire GDELT 1.0 Event database and compiles a timeline spanning the last four decades, listing the top five countries that experienced the greatest increase in their instability level each year. […]
In the aftermath of accusations a Facebook security employee was misusing his access to stalk women, it turns out the company actually sends alerts to employees when their profiles are accessed – what does this double standard tell us about how it views the privacy and security of its users' data? Read The Full Article.
Google’s new Duplex AI assistant opens up an entirely new world of conversational systems that can interact autonomously with humans to take action and gain information. Read The Full Article.
Social media platforms like Facebook are increasingly betting their futures on AI-powered moderation, but could those tools actually be making things worse? Read The Full Article.
GDPR's rollout is not only rolling back the prior protections its citizens enjoyed, but is firmly entrenching, under legal protection, all of the most privacy invasive elements of our modern digital surveillance society, including helping to roll out mass social media facial recognition. The article was later cited in an El Pais article. Read The […]
Last year Facebook tried reframing facial recognition from a privacy debate into a major security tool that would stop fraudulent impersonation accounts, but it turns out the tool’s design appears to limit it to being PR theatrics rather than an actual safety feature. Read The Full Article.
Even as Cambridge University's ethics panel rejected Kogan’s Facebook research for lacking informed consent, the academic community condemns Facebook’s new privacy policies that prioritize user safety and make the mass harvesting of private information that is at the heart of their work harder. Read The Full Article.
The story of a Facebook security engineer fired for allegedly stalking women online and potentially exploiting his privileged access to private user data reminds us that companies need to foster a culture of treating private data misuse as seriously as they treat media leaks. Read The Full Article.
As AI becomes the go-to tool of the online world, companies are reporting only their successes, while the public and government regulators are kept in the dark as to how often these algorithms make mistakes and, most importantly, if they are right more often than they are wrong. Read The Full Article.
Facebook quietly announced this week that its secret news outlet ranking is now live and being actively used to “boost” and “suppress” news but refuses to permit inspection of the list or how it is constructed. Read The Full Article.