SummAE: Zero-Shot Abstractive Text Summarization using Length-Agnostic Auto-Encoders
This work addresses the problem of generating abstractive summaries without task-specific training data for researchers and practitioners in NLP, representing an incremental advance by adapting existing auto-encoder methods to a new benchmark.
The authors tackled zero-shot abstractive text summarization of paragraphs by proposing SummAE, a neural model using length-agnostic auto-encoders, and introduced the ROCSumm benchmark based on ROCStories. Their model outperformed extractive baselines, with specific architectural choices and pre-training techniques significantly improving performance, though no concrete numbers were provided in the abstract.
We propose an end-to-end neural model for zero-shot abstractive text summarization of paragraphs, and introduce a benchmark task, ROCSumm, based on ROCStories, a subset for which we collected human summaries. In this task, five-sentence stories (paragraphs) are summarized with one sentence, using human summaries only for evaluation. We show results for extractive and human baselines to demonstrate a large abstractive gap in performance. Our model, SummAE, consists of a denoising auto-encoder that embeds sentences and paragraphs in a common space, from which either can be decoded. Summaries for paragraphs are generated by decoding a sentence from the paragraph representations. We find that traditional sequence-to-sequence auto-encoders fail to produce good summaries and describe how specific architectural choices and pre-training techniques can significantly improve performance, outperforming extractive baselines. The data, training, evaluation code, and best model weights are open-sourced.