Recently, transfer-learning by unsupervised pre-training and fine-tuning has shown great success on a number of tasks. The paucity of data for multi-document summarization (MDS) in the news domain, especially makes this approach practical. However, while existing literature mostly formulate unsupervised learning objectives tailored for/around the summarization problem we find that MDS can benefit directly from models pre-trained on other downstream supervised tasks such as sentence extraction, paraphrase generation and sentence compression. We carry out experiments to demonstrate the impact of zero-shot transfer-learning from these downstream tasks on MDS. Since it is challenging to train end-to-end encoder-decoder models on MDS due to i) the sheer length of the input documents, and ii) the paucity of training data. We hope this paper encourages more work on these downstream tasks as a means to mitigating the challenges in neural abstractive MDS.

We propose a system that improves performance on single document summarization task using the CNN/DailyMail and Newsroom datasets. It follows the popular encoder-decoder paradigm, but with an extra focus on the encoder. The intuition is that the probability of correctly decoding an information significantly lies in the pattern and correctness of the encoder. Hence we introduce, encode –encode – decode. A framework that encodes the source text first with a transformer, then a sequence-to-sequence (seq2seq) model. We find that the transformer and seq2seq model complement themselves adequately, making for a richer encoded vector representation. We also find that paying more attention to the vocabulary of target words during abstraction improves performance. We experiment our hypothesis and framework on the task of extractive and abstractive single document summarization and evaluate using the standard CNN/DailyMail dataset and the recently released Newsroom dataset.

Paraphrase generation aims to improve the clarity of a sentence by using different wording that convey similar meaning. For better quality of generated paraphrases, we propose a framework that combines the effectiveness of two models – transformer and sequence-to-sequence (seq2seq). We design a two-layer stack of encoders. The first layer is a transformer model containing 6 stacked identical layers with multi-head self attention, while the second-layer is a seq2seq model with gated recurrent units (GRU-RNN). The transformer encoder layer learns to capture long-term dependencies, together with syntactic and semantic properties of the input sentence. This rich vector representation learned by the transformer serves as input to the GRU-RNN encoder responsible for producing the state vector for decoding. Experimental results on two datasets-QUORA and MSCOCO using our framework, produces a new benchmark for paraphrase generation.

We demonstrate the viability of knowledge transfer between two related tasks: machine reading comprehension (MRC) and query-based text summarization. Using an MRC model trained on the SQuAD1.1 dataset as a core system component, we first build an extractive query-based summarizer. For better precision, this summarizer also compresses the output of the MRC model using a novel sentence compression technique. We further leverage pre-trained machine translation systems to abstract our extracted summaries. Our models achieve state-of-the-art results on the publicly available CNN/Daily Mail and Debatepedia datasets, and can serve as simple yet powerful baselines for future systems. We also hope that these results will encourage research on transfer learning from large MRC corpora to query-based summarization.