Mining Knowledge Graphs from Text

Knowledge Graph Construction From Text

AAAI 2017 Tutorial (schedule)

February 5, 2017

Location: Continental 7-9, Ballroom Level (Hilton SF Union Sq)

Jay Pujara, Sameer Singh, Bhavana Dalvi

With the proliferation of large collections of unstructured text, the problem of extracting structured knowledge and integrating it into a coherent knowledge graph has become increasingly important. Applications that rely on structured knowledge representations include digital assistants (Siri, Alexa, Cortana, and Google Now), question answering, summarization, and as well as many downstream autonomous decision-making. Due to its importance, this area has been an active area of research spanning areas of natural language processing, information extraction, information integration, databases, search, and machine learning.

Our goal is to present an accessible and structured overview of the existing approaches to extracting candidate facts from text and incorporating these into a well-formed knowledge graph. Our approach includes identifying the common themes and challenges in the area, and comparing and contrasting the existing approaches on the basis of these aspects. We believe such a unifying framework will provide the necessary tools and perspectives to enable the newcomers to the field to explore, evaluate, and develop novel techniques for automated knowledge graph construction.

Outline (with draft slides)

Part 1: Knowledge Graph Primer [ Slides ]

• What is a Knowledge Graph?
• Why are Knowledge Graphs Important?
• Where do Knowledge Graphs come from?
• Knowledge Representation Choices
• Problem Overview

Part 2: Knowledge Extraction from Text

• NLP Fundamentals [ Slides ]
  • Tokenization, chunking
  • Part-of-speech tagging
  • Named entity recognition
  • Dependency parsing
  • Entity resolution, coreference, and linking
• Information Extraction [ Slides ]
  • Defining knowledge domains
  • Learning knowledge extractors
  • Scoring extracted knowledge
  • Categories of IE techniques
  • Compositional models: Knowledge fusion
  • IE systems in practice

Coffee Break

Part 3: Knowledge Graph Construction

• Graph construction overview
• Probabilistic Models [ Slides ]
  • Motivation
  • Graphical models
  • Random walk approaches
• Embedding Techniques [ Slides ]
  • Relation extraction techniques
    • Matrix factorization
    • Embedding entity pairs
  • Graph completion techniques
    • Tensor factorization
    • Entity and relation embeddings
    • Compositional models

Part 4: Critical Overview and Conclusion [ Slides ]

• Summary
• Success stories
• Datasets, tasks, softwares
• Exciting active research
• Future research directions