Skip to content
conftrace_
2018 EMNLP EMNLP 2018

Interpretable Word Embedding Contextualization

Abstract

AbstractIn this paper, we propose a method of calibrating a word embedding, so that the semantic it conveys becomes more relevant to the context. Our method is novel because the output shows clearly which senses that were originally presented in a target word embedding become stronger or weaker. This is possible by utilizing the technique of using sparse coding to recover senses that comprises a word embedding.

🌉 Interdisciplinary Bridge - Artificial Intelligence and Natural Language Processing
🐣 Hot Topic Early Bird - word sense disambiguation

Authors

Kyoung-Rok Jang, Sung-Hyon Myaeng, Sang-Bum Kim

Topics

Machine Learning > Core Methods > Representation Learning Machine Learning > Core Methods > Embedding Learning Machine Learning > Optimization & Theory > Theory Natural Language Processing > Resources & Methods > Text Representation Deep Learning > Learning Types > Representation Learning Machine Learning > Core Methods > Sparse Coding

Keywords

sparse coding word sense disambiguation semantic analysis semantic representation word embedding sense disambiguation
Download PDF

More from EMNLP 2018

All EMNLP 2018 papers →
Language Models Learn POS First
Predicting and interpreting embeddings for out of vocabulary words in downstream tasks
Probing sentence embeddings for structure-dependent tense
State Gradients for RNN Memory Analysis
Extracting Syntactic Trees from Transformer Encoder Self-Attentions
Portable, layer-wise task performance monitoring for NLP models