Alexander M. Rush, Sumit Chopra, Jason Weston
Summarization based on text extraction is inherently limited, but generation-style abstractive methods have proven challenging to build. In this work, we propose a fully data-driven approach to abstractive sentence summarization. Our method utilizes a local attention-based model that generates each word of the summary conditioned on the input sentence. While the model is structurally simple, it can easily be trained end-to-end and scales to a large amount of training data. The model shows significant performance gains on the DUC-2004 shared task compared with several strong baselines.
| Task | Dataset | Metric | Value | Model |
|---|---|---|---|---|
| Text Summarization | DUC 2004 Task 1 | ROUGE-1 | 28.18 | Abs+ |
| Text Summarization | DUC 2004 Task 1 | ROUGE-2 | 8.49 | Abs+ |
| Text Summarization | DUC 2004 Task 1 | ROUGE-L | 23.81 | Abs+ |
| Text Summarization | DUC 2004 Task 1 | ROUGE-L | 22.05 | ABS |
| Text Summarization | GigaWord | ROUGE-1 | 31 | Abs+ |
| Text Summarization | GigaWord | ROUGE-1 | 30.88 | Abs |
| Text Summarization | DUC 2004 Task 1 | ROUGE-1 | 26.55 | Abs |
| Text Summarization | DUC 2004 Task 1 | ROUGE-2 | 7.06 | Abs |
| Text Summarization | DUC 2004 Task 1 | ROUGE-L | 22.05 | Abs |
| Extractive Text Summarization | DUC 2004 Task 1 | ROUGE-1 | 26.55 | Abs |
| Extractive Text Summarization | DUC 2004 Task 1 | ROUGE-2 | 7.06 | Abs |
| Extractive Text Summarization | DUC 2004 Task 1 | ROUGE-L | 22.05 | Abs |