An Inside Look at Knowledge Graph

Jon Mitchell recently interviewed Emily Moxley, the lead product manager of Google's Knowledge Graph. Asked about the goal of Knowledge Graph, Moxley responded, "It’s about mapping the real world into something that computers can understand. It’s all about taking real-world objects, understanding what those things are and what the important things about them are, and also their relationship to other things. That’s the part I find the most fascinating: What is the relationship between all these different things?"

Mitchell also asked about how the Knowledge Graph is maintained: "It’s very actively maintained by Google employees. Metaweb, before, was this repository of entities and facts, and [the company's employees] were very much using their intuition about what people cared about and what information to go find. Since they’ve been acquired by Google, Google has all these users looking for information, and from that, we’re able to see what things are interesting about the world. Through that, we’re able to grow the Knowledge Graph in an efficient direction. We usually find that the users’ interests match our intuition, but there are serendipitous things that wouldn’t necessarily match your intuitions. Google knows that people who query for one thing also query for this other thing. So you see some interesting relationships when you look at aggregated user sessions. One of the next steps is actually explaining those serendipitous relationships. So what is it about this collection of actors that’s similar among them? What is it about this movie that explains why people often search for these five movies together?"

Read more here.

Image: Courtesy Google