Shashi Narayan

Researcher at Google specializing in natural language generation, abstractive summarization, and evaluation methods for generated text. Developed the Topic-aware Convolutional Seq2Seq (TConvS2S) model used in summarization research.

Sources in this wiki

• 2020-maynez-faithfulness-factuality-summarization

Topics

• abstractive-summarization, natural-language-generation, hallucination

Notes

Co-author on work investigating hallucinations in abstractive summarization and the evaluation of faithfulness and factuality in neural text generation.