Integrating Distributional Lexical Contrast into Word Embeddings for Antonym-Synonym Distinction

  title={Integrating Distributional Lexical Contrast into Word Embeddings for Antonym-Synonym Distinction},
  author={K. A. Nguyen and Sabine Schulte im Walde and Ngoc Thang Vu},
  • K. A. Nguyen, Sabine Schulte im Walde, Ngoc Thang Vu
  • Published 2016
  • Computer Science
  • ArXiv
  • We propose a novel vector representation that integrates lexical contrast into distributional vectors and strengthens the most salient features for determining degrees of word similarity. The improved vectors significantly outperform standard models and distinguish antonyms from synonyms with an average precision of 0.66-0.76 across word classes (adjectives, nouns, verbs). Moreover, we integrate the lexical contrast vectors into the objective function of a skip-gram model. The novel embedding… CONTINUE READING
    94 Citations

    Figures, Tables, and Topics from this paper

    Distinguishing antonymy, synonymy and hypernymy with distributional and distributed vector representations and neural networks
    • PDF
    Revisit Word Embeddings with Semantic Lexicons for Modeling Lexical Contrast
    • 7
    Injecting Lexical Contrast into Word Vectors by Guiding Vector Space Specialisation
    • 7
    • PDF
    Learning Lexical Subspaces in a Distributional Vector Space
    • 3
    Synonyms and Antonyms: Embedded Conflict
    • 1
    • PDF
    Explicit Retrofitting of Distributional Word Vectors
    • 52
    • Highly Influenced
    • PDF
    Specializing Distributional Vectors of All Words for Lexical Entailment
    • 12
    • PDF


    A Multitask Objective to Inject Lexical Contrast into Distributional Semantics
    • 36
    • PDF
    Word Embedding-based Antonym Detection using Thesauri and Distributional Information
    • 84
    • PDF
    Evaluating WordNet-based Measures of Lexical Semantic Relatedness
    • 1,495
    • PDF
    Taking Antonymy Mask off in Vector Space
    • 28
    • PDF
    Efficient Estimation of Word Representations in Vector Space
    • 16,728
    • Highly Influential
    • PDF
    WordNet: a lexical database for English
    • 12,346
    • Highly Influential
    • PDF