Bayesian Paragraph Vectors
This work provides a probabilistic framework for text representation, offering incremental improvements in NLP tasks like sentiment analysis and paraphrase detection.
The authors tackled the problem of representing variable-length text pieces by proposing a Bayesian interpretation of paragraph vectors, which improved performance in sentiment analysis and paraphrase detection by incorporating posterior uncertainty.
Word2vec (Mikolov et al., 2013) has proven to be successful in natural language processing by capturing the semantic relationships between different words. Built on top of single-word embeddings, paragraph vectors (Le and Mikolov, 2014) find fixed-length representations for pieces of text with arbitrary lengths, such as documents, paragraphs, and sentences. In this work, we propose a novel interpretation for neural-network-based paragraph vectors by developing an unsupervised generative model whose maximum likelihood solution corresponds to traditional paragraph vectors. This probabilistic formulation allows us to go beyond point estimates of parameters and to perform Bayesian posterior inference. We find that the entropy of paragraph vectors decreases with the length of documents, and that information about posterior uncertainty improves performance in supervised learning tasks such as sentiment analysis and paraphrase detection.