Interpretable Word Embeddings via Informative Priors
arXiv:1909.01459
Abstract
Word embeddings have demonstrated strong performance on NLP tasks. However, lack of interpretability and the unsupervised nature of word embeddings have limited their use within computational social science and digital humanities. We propose the use of informative priors to create interpretable and domain-informed dimensions for probabilistic word embeddings. Experimental results show that sensible priors can capture latent semantic concepts better than or on-par with the current state of the art, while retaining the simplicity and generalizability of using priors.
arXiv comment10 pages, 2 figures, EMNLP 2019
References in corpus (7)
- Efficient Estimation of Word Representations in Vector Space
- Distributed Representations of Words and Phrases and their Compositionality
- Word Embeddings Quantify 100 Years of Gender and Ethnic Stereotypes
- Word Representations via Gaussian Embedding
- Exponential Family Embeddings
- Bayesian Neural Word Embedding
- Dynamic Bernoulli Embeddings for Language Evolution