Title: Creating the Knowledge Graph for SIRI

Speaker: Xiao Ling, Apple

Abstract

At Apple, one of the goals Siri aims at is to be a know-it-all question answering system, capable of answering questions from hundreds of millions of users about nearly anything. The question answering system is backed by a knowledge graph that was automatically constructed from a vast number of data sources including natural language text, HTML tables, and many others.

In this talk, we will give a brief overview of automatic knowledge base construction (AKBC) at Siri and discuss two concrete problems, Wikipedia infobox extraction and entity resolution. Wikipedia infobox extraction has been somewhat considered solved in the literature. We will show why this is still a challenging problem in the AKBC context and present our work published in NAACL 2019. We will also discuss in-progress work in tackling entity resolution by embedding entities to vector spaces for matching and promising results in our preliminary experiments.

Bio

Xiao Ling is a research engineer at Apple Siri Knowledge. He received his PhD in Computer Science and Engineering from the University of Washington in 2015. His work focuses on Information Extraction, Natural Language Processing and Machine Learning. He was an early engineer at Lattice Data Inc., which was acquired by Apple in 2017.