Knowledge gathering, representation, and reasoning are at the heart of challenges of Artificial Intelligence. Extracting knowledge buried in the text and integrating it into a knowledge base are one step towards such challenges. Moreover, the curated incomplete knowledge base can be populated by inferring new facts from existing ones. Recent years have seen significant advances in both academia and industry of knowledge base area with the rise of deep learning.
In this talk, I will discuss recent techniques used for constructing a knowledge base from text. This includes methods such as entity recognition and relation extraction that directly extract knowledge of entities and relations from corpus. Then I will cover knowledge base completion methods that infer new facts from the existing knowledge base.