Lecture 16 Dependency Grammar

Dependency Grammar

Dependency Grammar

• Dependency grammar offers a simpler approach to CFG:

  • Describe relations between pairs of words
  • Namely, between heads and dependents

• Deal better with languages that are morphologically rich and have a relatively free word order

  • CFG need a separate rule for each possible place a phrase can occur in

• Head-dependent relations similar to semantic relations between words

  • More useful for applications: coreference resolution, information extraction

Dependency Relations

• Captures the grammatical relation between:

  • Head: central word
  • Dependent: supporting word

• Grammatical relation: subject, direct object, …

• Many dependency theories and taxonomies proposed for different languages

• Universal dependency: a framework to create a set of dependency relations that are computationally useful and cross-lingual

  

Application: Question Answering

• Dependency tree more directly represents the core of the sentence: who did want to whom?
  • Captured by the links incident on verb nodes

  

Application: Information Extraction

• Brasilia, the Brazilian capital, was founded in 1960. to capital(Brazil, Brasilia); founded(Brasilia, 1960)
• Dependency tree captures relations succinctly

  

Dependency vs. Constituency

• Dependency tree:

  • Each node is a word token
  • One node is chosen as the root
  • Directed edges link heads and their dependents

• Constituency tree:

  • Forms a hierarchical tree
  • Word tokens are the leaves
  • Internal nodes are constituent phrases

• Both use POS

Properties of a Dependency Tree

• Each word has a single head (parent)
• There is a single root node
• There is a unique path to each word from the root
• All arcs should be projective

Projectivity

• An arc is projetive if there is a path from head to every word that lies between head and the dependent

  

• Dependency tree is projective if all arcs are projective. In other words, a dependency tree is projective if it can be drawn with no crossing edges

• Most sentences are projective, but exception exist

Treebank Conversion

• A few dependency treebanks but many constituency treebanks
• Some constituency treebanks can be converted into dependencies
• Dependency trees generated from constituency trees are always projective
• Main idea: Identify head-dependent relations in constituency structure and the corresponding dependency relations
  • Use various heuristics
  • Often with manual correction

Transition-based Parsing

Dependency Parsing

• Find the best structure for a given input sentence

• Two main approaches:

  • Transition-based: Bottom-up greedy method
  • Graph-based: encodes problem using nodes/edges and use graph theory methods to find optimal solutions

• Caveat:

  • Transition-based parsers can only handle projective dependency trees
  • Less applicable for languages where cross-dependencies are common

Transition-Based Parsing

• Processes word from left to right

• Maintain two data structures:

  • Buffer: Input words yet to be processed
  • Stack: Store words that are being processed

  

• At each step, perform one of the 3 actions:

  • Shift: Move a word from buffer to stack
  • Left-Arc: Assign current words as the head of the previous word in stack
  • Right-Arc: Assign previous word as head of current word in stack

• E.g.

  
  
  
  

• For simplicity, omit labels on the dependency relations. In practice, we parameterize the left-arc and right-arc with dependency labels:

  • E.g. left-arc-nsubj or right-arc-dobj

• Expands the list of actions to > 3 types

• Assume an oracle that tells the correct action at every step. Given a dependency tree, the role of oracle is to generate a sequence of ground truth action.

Parsing Model

• Train a supervised model to mimic the actions of the oracle

  • To learn at every step the correct action to take
  • At test time, the trained model can be used to parse a sentence to create the dependency tree

• Parse as Classification:

  • Input:
    • Stack: top two elements: s₁ and s₂
    • Buffer: first element: b₁
  • Output:
    • 3 classes: shift, left-arc, right-arc
  • Features:
    • word, part-of-speech

• Traditionally SVM works best. Nowadays, deep learning models are SOTA

• Weakness: local classifier based on greedy search

• Solution:

  • Beam Search: Keep track of top-N best actions
  • Dynamic oracle: during training, use predicted actions occasionally
  • Graph-based parser

Graph-based Parsing

Graph-based Parsing

• Given an input sentence, construct a fully-connected, weighted, directed graph

• Vertices: all words

• Edges: head-dependent arcs

• Weight: score based on training data (relation that is frequently observed receive in a higher score)

• Objective: find the maximum spanning tree (Kruskal’s algorithm)

• Advantage

  • Can produce non-projective trees

  • Score the entire trees:

    • Avoid making greedy local decisions like transition-based parsers
    • Captures long dependencies better

• E.g.:

  

• Caveat: Tree may contain cycles

  • Solution: Need to do cleanup to remove cycles