Overview
Explore the fundamentals of syntax and parsing in multilingual natural language processing through this comprehensive lecture by Graham Neubig and Aditi Chaudhary. Delve into major word order concepts, dependency parsing models, and gain insights into AutoLex, a system for linguistic discovery. Learn about Universal Dependencies, the distinctions between syntax and semantics, and various parsing techniques including shift-reduce, graph-based, and transition-based approaches. Examine dynamic programming, feature extraction, and the challenges of dependency parsing across languages. Discover linguistic-informed constraints, agreement requirements, and prediction tasks in NLP. Gain practical knowledge of NLP toolkits and frameworks applicable to multilingual contexts.
Syllabus
Introduction
Universal Dependencies
Syntax vs Semantic
Semantics
Dependencies
Parsing
ShiftReduce
Classification
Graphbased
Graphbased vs Transitionbased
Dynamic Programming
Feature Extractor
Difficulty
Order insensitive encoder
Dependency parsing
Linguistic informed constraints
Definition of required agreement
Prediction task
Linguistic questions
General framework
Toolkit
Summary
Taught by
Graham Neubig