# Word Embeddings: A Survey

**Authors:** Felipe Almeida, Geraldo Xexéo  
**arXiv:** [1901.09069](https://arxiv.org/abs/1901.09069)  
**Date:** 2019-01-25  
**Summary page:** https://www.summarizepaper.com/en/paper/1901.09069v1/

## Key points

- Word embeddings are fixed-length, dense, and distributed representations for words that encode significant syntactic and semantic information.
- They are built on the distributional hypothesis and have been proven to be useful in various downstream natural language processing (NLP) tasks such as sentiment analysis, named entity recognition, and machine translation.
- Various techniques used to create these embeddings include neural network-based approaches such as skip-gram and CBOW models as well as matrix factorization methods like GloVe.
- Challenges associated with building word embeddings include handling out-of-vocabulary words and dealing with polysemy.
- This survey provides a comprehensive overview of the current state-of-the-art techniques for constructing word embeddings and their practical applications in NLP.

## Summary

In their paper titled "Word Embeddings: A Survey," authors Felipe Almeida and Geraldo Xexéo provide an in-depth analysis of the recent strategies for constructing word embeddings. These fixed-length, dense, and distributed representations for words are built on the distributional hypothesis and have been proven to encode significant syntactic and semantic information. The authors discuss various techniques used to create these embeddings, including neural network-based approaches such as skip-gram and CBOW models as well as matrix factorization methods like GloVe. They also explore the challenges associated with building word embeddings, such as handling out-of-vocabulary words and dealing with polysemy. Furthermore, the authors highlight the usefulness of word embeddings in various downstream natural language processing (NLP) tasks such as sentiment analysis, named entity recognition, and machine translation. This survey provides a comprehensive overview of the current state-of-the-art techniques for constructing word embeddings and their practical applications in NLP. Overall, this survey offers a thorough understanding of how to effectively use word embedding techniques in NLP tasks.