Follow Datanami:
September 17, 2020

New AI Tool Maps the Families of the Bible, A Song of Ice and Fire

Diffbot is obsessed with knowledge. The Palo Alto-based knowledge-as-a-service company is on a quest to “build the first comprehensive map of all human knowledge,” according to its founder and CEO Mike Tung – and it’s doing that using a medley of machine learning methods to automatically crawl, index, and correlate the content of the internet. Now, Diffbot is announcing the launch of the latest weapon in its gnostic arsenal: a new natural language processing (NLP) API that is fully integrated into Diffbot’s multilingual knowledge graph.

To showcase the newly available deep learning system, Diffbot turned to two sets of famous texts: the Bible… and A Song of Ice and Fire, the series that served as the basis for the hit show Game of Thrones. While those works may seem worlds apart, they share a common trait of great interest to Diffbot: complex genealogies. The winding and weaving family webs of both the Bible and the fantasy world of Westeros pose a challenge to even astute readers – and so, they posed the perfect challenge for Diffbot’s NLP API. 

Diffbot’s API, which uses its vast knowledge of the web to analyze free text, chewed on the two massive documents, spitting out full (and accurate) genealogies for all traceable figures mentioned in both. For A Song of Ice and Fire, Diffbot even produced an infographic mapping out the complex familial ties in the series and highlighting some of the impressive statistics: 3,663 characters detected, 855 parent-child relationships, 700 sibling relationships, 322 spousal relationships – and 15 minutes of NLP to do all of it.

An infographic displaying stylized results from Diffbot’s NLP analysis of A Song of Ice and Fire. Image courtesy of Diffbot.

“With this new breakthrough, Diffbot continues to push the boundaries in achieving its vision in building an ImageNet for human language and realizing a Semantic Web that will transform the business world,” Tung said. “We have been testing its capabilities in beta and the results have been extremely promising in illustrating the true impact it will have on the professional world, especially in the fields of science, history, education, and religion.” 

Diffbot provides its knowledge graph for free to researchers, but the company says it has around 400 paying customers, as well – including some big names like Adobe, Cisco, eBay, Salesforce, and Samsung. 

“I believe we are creating a system that will serve as a dedicated enterprise search engine,” Tung continued. “The real Semantic Web – unbound by the ad-support model where businesses in every industry can find accurate and extensive information and build customized workflows on top of their queries.”