Abstractive Summarization using Variational Autoencoders
Natural Language Processing
Present summarization techniques fail for long documents and hallucinate facts. Exploring methods to generate a faithful abstractive summary by editing facts extracted from the document using variational inference.