Google Uses Open Information Retrieval to Guess Unknown Entities in Previously Unseen Content

Open Information Retrieval? What is that? Why is it so important? In this clip, Koray Gübür explains that Open Information Retrieval is designed to define and understand unknown entities without the Knowledge Graph. This insight is phenomenally important, and often vastly underestimated. - Kalicube Knowledge Nuggets

Transcript:

It is open information retrieval. In open information retrieval, they try to find unknown entities. So let's say we have a document, completely new document, there is no information, and we can even create a random document there and they will use open information instruction there for creating a graph just from that web document and anything that they explored there will be clustered with other teams. So it might not exist directly within the Knowledge Graph, but there will be also unknown entities as well.

Right. Which explains two things. One of which is at the NLP in Kalicube Pro extracts entities. And I've got the difference between recognized and guessed ones. And those are the guessed ones. And the other is I was talking to Ali Alvi from Bing who does the Featured Snippet, the Q and A for Bing. And he was saying, "Actually, we often don't use the Knowledge Graph." (Presumably, they're just doing exactly what you're talking about there)And it's taken me two years for anybody to actually explain that to me.

It's okay.

"Kalicube Tuesdays is hosted by Jason Barnard"