Inducting Domain-specific Polarity Lexica

Student: N/A

Supervisor: Manfred Klenner

Introduction

Certain adjectives (and also verbs) prefer to combine with positive (or negative) nouns, thus, they have a polarity preference. These can be learned by means of a large text corpus and a polarity lexicon. Applied to domain-specific texts, domain-specific polarity nouns can be identified. This works amazingly well and is helpful.

Aim and purpose

Automic induction of polarity nouns in various domains (Financial sector, pharmaceuticals, etc.)

Requirements

  • Programming knowledge in Python
  • Course in quantitative methods in Computational Linguistics

Procedure

  • Learning polarity preferences: extraction of the adjectives from the body of the text that modify nouns from the lexicon
  • Estimate preferences via maximum likelihood
  • Apply to domain corpus
  • Evaluate

Necessary resources

  • Polarity lexicon is available
  • Text corpus for German is available

Literature

Klenner, Manfred / Amsler, Michael / Hollenstein, Nora (2014): Inducing Domain-specific Noun Polarity Guided by Domain-independent Polarity Preferences of Adjectives. In: Proceedings of the 5th Workshop on Computational Approaches to Subjectivity, Sentiment and Social Media Analysis. Pages 18-23.