Syntactically Look-Ahead Attention Network for Sentence Compression
This work addresses sentence compression for NLP applications, offering an incremental improvement by enhancing informativeness and readability in generated summaries.
The paper tackled the problem of generating ungrammatical sentences in sequence-to-sequence models for sentence compression by proposing SLAHAN, a model that explicitly tracks dependency relationships during decoding, achieving best scores of 85.5 F1, 79.3 ROUGE-1, 71.3 ROUGE-2, and 79.1 ROUGE-L on the Google dataset.
Sentence compression is the task of compressing a long sentence into a short one by deleting redundant words. In sequence-to-sequence (Seq2Seq) based models, the decoder unidirectionally decides to retain or delete words. Thus, it cannot usually explicitly capture the relationships between decoded words and unseen words that will be decoded in the future time steps. Therefore, to avoid generating ungrammatical sentences, the decoder sometimes drops important words in compressing sentences. To solve this problem, we propose a novel Seq2Seq model, syntactically look-ahead attention network (SLAHAN), that can generate informative summaries by explicitly tracking both dependency parent and child words during decoding and capturing important words that will be decoded in the future. The results of the automatic evaluation on the Google sentence compression dataset showed that SLAHAN achieved the best kept-token-based-F1, ROUGE-1, ROUGE-2 and ROUGE-L scores of 85.5, 79.3, 71.3 and 79.1, respectively. SLAHAN also improved the summarization performance on longer sentences. Furthermore, in the human evaluation, SLAHAN improved informativeness without losing readability.