Header menu link for other important links
X
Leveraging ontological knowledge for neural language models
Ameet Deshpande, Monisha Jegadeesan
Published in Association for Computing Machinery
2019
Pages: 350 - 353
Abstract
Neural Language Models such as Word2Vec and GloVe have been shown to encode semantic relatedness between words. Improvements in unearthing these embeddings can ameliorate performance in numerous downstream applications such as sentiment analysis, question answering, and dialogue generation. Lexical ontologies such as WordNet are known to supply information about semantic similarity rather than relatedness. Further, extracting word embeddings from small corpora is daunting for data-hungry neural networks. This work shows how methods that conflate Word2Vec and Ontologies can achieve better performance, reduce training time and help adapt to domains with a minimum amount of data. © 2019 Copyright held by the owner/author(s).
About the journal
JournalData powered by TypesetACM International Conference Proceeding Series
PublisherData powered by TypesetAssociation for Computing Machinery
Open AccessNo
Concepts (14)
  •  related image
    Computational linguistics
  •  related image
    Embeddings
  •  related image
    Knowledge management
  •  related image
    Semantics
  •  related image
    Sentiment analysis
  •  related image
    DIALOGUE GENERATIONS
  •  related image
    DOMAIN TRANSFERS
  •  related image
    DOWNSTREAM APPLICATIONS
  •  related image
    Hierarchy
  •  related image
    Question answering
  •  related image
    Semantic relatedness
  •  related image
    Semantic similarity
  •  related image
    WORD VECTORS
  •  related image
    Ontology