Constituency-based Hyponymy Extraction COMP 762 Chianyu Liu, - - PowerPoint PPT Presentation

Constituency-based Hyponymy Extraction

COMP 762 Chianyu Liu, 260576898

Hyponym and Hypernym

• Describes a “type-of” relationship
• E.g.
• Meronym: describes a part-whole relationship
• Homonym: a word that has two unrelated meanings
• Polyseme: a word with two related meanings

Constituency-based parse tree

• Ordered, rooted tree that represents the syntactic structure of a text
• Breaks a sentence, S, into sub-phrases (e.g. NP, VP, PP etc.)

○ Nodes are phrases ○ Leafs are words ○ Edges are unlabeled

• Vs. Dependency-based parse tree
• http://nlp.stanford.edu:8080/parser/index.jsp

Tregex

• An utility for identifying patterns in trees
• Like regular expressions for strings
• Use symbols to denote relations

○ A < B: A is the parent of B ○ A << B: A is an ancestor of B ○ A$B: A and B are siblings

Tregex Example

Pattern Matching

• Hyponymy categories defined in the paper

Pattern Description Example HKO Hypernym -> Keywords -> Hyponym …, such as … OKH Hyponym -> Keyworks -> Hypernym … are considered as … HO Hyponym -> Hyponym Section header KHO Keywords -> Hypernym -> Hyponym Following types of …

WordNet

• A large lexical database of English
• Nouns, verbs, adjectives and adverbs are grouped into sets of synsets
• Synsets are interlinked by means of conceptual-semantic and lexical relations
• The most frequently encoded relation among synsets is the

super-subordinate relation (i.e. hypernym and hyponym)

• http://wordnetweb.princeton.edu/perl/webwn

Ontology

• A specification of a conceptualization
• Describes the representation and relationships that can exist for entities

(objects, properties, etc.) in a particular domain

• Ontology is a standard to represent knowledge, and enables knowledge to be

shared and reused

References

• G. Andrew, “The Wonderful World of Tregex,” in Stanford: Natural Language Processing.
• M. C. Evans, J. Bhatia, S. Wadkar, and T. D. Breaux, “An Evaluation of Constituency-Based Hyponymy Extraction from Privacy

Policies,” 2017 IEEE 25th International Requirements Engineering Conference (RE), 2017.

• P. University, “What is WordNet?,” Princeton University, 17-Mar-2015. [Online]. Available: https://wordnet.princeton.edu/.
• S. Gole, “Part of speech tagging using OpenNLP,” Sager Gole's Blog, 18-Jun-2015.