Knowledge Graphs

Chapter 3. Data Management for Actionable Knowledge

When the data team of the International Consortium of Investigative Journalists (ICIJ) received a data dump that came to be known as the Panama Papers, they probably thought they’d be looking for the digital equivalent of a needle in a haystack.¹

With such a large and complex set of data, it may have seemed an insurmountable task for a small team of knowledge workers, but their decision to build a knowledge graph with the data fundamentally changed that situation. The knowledge graph the ICIJ built gave context and connections to the data. The complex, multiyear, multimedia data in the knowledge graph was linked in a way that investigators just needed to follow the connections to uncover often scandalous stories that we have been reading in the news since 2015.

In unraveling the Panama Papers, the ICIJ demonstrated that explicit connections in data are transformative for consumers, whether they are human or software agents. Amongst the in excess of 214,000 entities and 140 prominent individuals discovered in the leak, the files also showed how major banks played a significant role in helping to move wealth offshore.

But that was not the end of the story. The work of the ICIJ is ongoing, and so the data team at ICIJ kept improving the data pipeline for populating the knowledge graph. They delivered a freshly updated graph to the journalists every day, despite ongoing changes like new code releases, additional data sources, and ...