A major challenge for scaling machine learning is training models to perform tasks that are very difficult or time-consuming for humans to evaluate. We present progress on this problem on the task of abstractive summarization of entire fiction novels. Our method combines learning from human feedback with recursive task decomposition: we use models trained on smaller parts of the task to assist humans in giving feedback on the broader task. We collect a large volume of demonstrations and comparisons from human labelers, and fine-tune GPT-3 using behavioral cloning and reward modeling to do summarization recursively. At inference time, the model first summarizes small sections of the book and then recursively summarizes these summaries to produce a summary of the entire book. Our human labelers are able to supervise and evaluate the models quickly, despite not having read the entire books themselves. Our resulting model generates sensible summaries of entire books, even matching the quality of human-written summaries in a few cases ($\sim5\%$ of books). We achieve state-of-the-art results on the recent BookSum dataset for book-length summarization. A zero-shot question-answering model using these summaries achieves state-of-the-art results on the challenging NarrativeQA benchmark for answering questions about books and movie scripts. We release datasets of samples from our model.
State-of-the-art abstractive summarization systems often generate \emph{hallucinations}; i.e., content that is not directly inferable from the source text. Despite being assumed incorrect, we find that much hallucinated content is factual, namely consistent with world knowledge. These factual hallucinations can be beneficial in a summary by providing useful background information. In this work, we propose a novel detection approach that separates factual from non-factual hallucinations of entities. Our method utilizes an entity's prior and posterior probabilities according to pre-trained and finetuned masked language models, respectively. Empirical results suggest that our approach vastly outperforms two baselines %in both accuracy and F1 scores and strongly correlates with human judgments. % on factuality classification tasks. Furthermore, we show that our detector, when used as a reward signal in an off-line reinforcement learning (RL) algorithm, significantly improves the factuality of summaries while maintaining the level of abstractiveness.
We present BARTpho with two versions, BARTpho-syllable and BARTpho-word, which are the first public large-scale monolingual sequence-to-sequence models pre-trained for Vietnamese. BARTpho uses the "large" architecture and the pre-training scheme of the sequence-to-sequence denoising autoencoder BART, thus it is especially suitable for generative NLP tasks. We conduct experiments to compare our BARTpho with its competitor mBART on a downstream task of Vietnamese text summarization and show that: in both automatic and human evaluations, BARTpho outperforms the strong baseline mBART and improves the state-of-the-art. We further evaluate and compare BARTpho and mBART on the Vietnamese capitalization and punctuation restoration tasks and also find that BARTpho is more effective than mBART on these two tasks. We publicly release BARTpho to facilitate future research and applications of generative Vietnamese NLP tasks. Our BARTpho models are available at https://github.com/VinAIResearch/BARTpho
We study generating abstractive summaries that are faithful and factually consistent with the given articles. A novel contrastive learning formulation is presented, which leverages both reference summaries, as positive training data, and automatically generated erroneous summaries, as negative training data, to train summarization systems that are better at distinguishing between them. We further design four types of strategies for creating negative samples, to resemble errors made commonly by two state-of-the-art models, BART and PEGASUS, found in our new human annotations of summary errors. Experiments on XSum and CNN/Daily Mail show that our contrastive learning framework is robust across datasets and models. It consistently produces more factual summaries than strong comparisons with post error correction, entailment-based reranking, and unlikelihood training, according to QA-based factuality evaluation. Human judges echo the observation and find that our model summaries correct more errors.
We introduce TVStoryGen, a story generation dataset that requires generating detailed TV show episode recaps from a brief summary and a set of documents describing the characters involved. Unlike other story generation datasets, TVStoryGen contains stories that are authored by professional screen-writers and that feature complex interactions among multiple characters. Generating stories in TVStoryGen requires drawing relevant information from the lengthy provided documents about characters based on the brief summary. In addition, we propose to train reverse models on our dataset for evaluating the faithfulness of generated stories. We create TVStoryGen from fan-contributed websites, which allows us to collect 26k episode recaps with 1868.7 tokens on average. Empirically, we take a hierarchical story generation approach and find that the neural model that uses oracle content selectors for character descriptions demonstrates the best performance on automatic metrics, showing the potential of our dataset to inspire future research on story generation with constraints. Qualitative analysis shows that the best-performing model sometimes generates content that is unfaithful to the short summaries, suggesting promising directions for future work.
This paper explores three simple data manipulation techniques (synthesis, augmentation, curriculum) for improving abstractive summarization models without the need for any additional data. We introduce a method of data synthesis with paraphrasing, a data augmentation technique with sample mixing, and curriculum learning with two new difficulty metrics based on specificity and abstractiveness. We conduct experiments to show that these three techniques can help improve abstractive summarization across two summarization models and two different small datasets. Furthermore, we show that these techniques can improve performance when applied in isolation and when combined.
This paper explores the effect of using multitask learning for abstractive summarization in the context of small training corpora. In particular, we incorporate four different tasks (extractive summarization, language modeling, concept detection, and paraphrase detection) both individually and in combination, with the goal of enhancing the target task of abstractive summarization via multitask learning. We show that for many task combinations, a model trained in a multitask setting outperforms a model trained only for abstractive summarization, with no additional summarization data introduced. Additionally, we do a comprehensive search and find that certain tasks (e.g. paraphrase detection) consistently benefit abstractive summarization, not only when combined with other tasks but also when using different architectures and training corpora.
Existing summarization systems mostly generate summaries purely relying on the content of the source document. However, even for humans, we usually need some references or exemplars to help us fully understand the source document and write summaries in a particular format. But how to find the high-quality exemplars and incorporate them into summarization systems is still challenging and worth exploring. In this paper, we propose RetrievalSum, a novel retrieval enhanced abstractive summarization framework consisting of a dense Retriever and a Summarizer. At first, several closely related exemplars are retrieved as supplementary input to help the generation model understand the text more comprehensively. Furthermore, retrieved exemplars can also play a role in guiding the model to capture the writing style of a specific corpus. We validate our method on a wide range of summarization datasets across multiple domains and two backbone models: BERT and BART. Results show that our framework obtains significant improvement by 1.38~4.66 in ROUGE-1 score when compared with the powerful pre-trained models, and achieve new state-of-the-art on BillSum. Human evaluation demonstrates that our retrieval enhanced model can better capture the domain-specific writing style.
Previous abstractive methods apply sequence-to-sequence structures to generate summary without a module to assist the system to detect vital mentions and relationships within a document. To address this problem, we utilize semantic graph to boost the generation performance. Firstly, we extract important entities from each document and then establish a graph inspired by the idea of distant supervision \citep{mintz-etal-2009-distant}. Then, we combine a Bi-LSTM with a graph encoder to obtain the representation of each graph node. A novel neural decoder is presented to leverage the information of such entity graphs. Automatic and human evaluations show the effectiveness of our technique.
With the rapid increase of multimedia data, a large body of literature has emerged to work on multimodal summarization, the majority of which target at refining salient information from textual and visual modalities to output a pictorial summary with the most relevant images. Existing methods mostly focus on either extractive or abstractive summarization and rely on qualified image captions to build image references. We are the first to propose a Unified framework for Multimodal Summarization grounding on BART, UniMS, that integrates extractive and abstractive objectives, as well as selecting the image output. Specially, we adopt knowledge distillation from a vision-language pretrained model to improve image selection, which avoids any requirement on the existence and quality of image captions. Besides, we introduce a visual guided decoder to better integrate textual and visual modalities in guiding abstractive text generation. Results show that our best model achieves a new state-of-the-art result on a large-scale benchmark dataset. The newly involved extractive objective as well as the knowledge distillation technique are proven to bring a noticeable improvement to the multimodal summarization task.
Abstractive text summarization is one of the areas influenced by the emergence of pre-trained language models. Current pre-training works in abstractive summarization give more points to the summaries with more words in common with the main text and pay less attention to the semantic similarity between generated sentences and the original document. We propose ARMAN, a Transformer-based encoder-decoder model pre-trained with three novel objectives to address this issue. In ARMAN, salient sentences from a document are selected according to a modified semantic score to be masked and form a pseudo summary. To summarize more accurately and similar to human writing patterns, we applied modified sentence reordering. We evaluated our proposed models on six downstream Persian summarization tasks. Experimental results show that our proposed model achieves state-of-the-art performance on all six summarization tasks measured by ROUGE and BERTScore. Our models also outperform prior works in textual entailment, question paraphrasing, and multiple choice question answering. Finally, we established a human evaluation and show that using the semantic score significantly improves summarization results.
Transformer models have achieved state-of-the-art results in a wide range of NLP tasks including summarization. Training and inference using large transformer models can be computationally expensive. Previous work has focused on one important bottleneck, the quadratic self-attention mechanism in the encoder. Modified encoder architectures such as LED or LoBART use local attention patterns to address this problem for summarization. In contrast, this work focuses on the transformer's encoder-decoder attention mechanism. The cost of this attention becomes more significant in inference or training approaches that require model-generated histories. First, we examine the complexity of the encoder-decoder attention. We demonstrate empirically that there is a sparse sentence structure in document summarization that can be exploited by constraining the attention mechanism to a subset of input sentences, whilst maintaining system performance. Second, we propose a modified architecture that selects the subset of sentences to constrain the encoder-decoder attention. Experiments are carried out on abstractive summarization tasks, including CNN/DailyMail, XSum, Spotify Podcast, and arXiv.
Contrastive learning models have achieved great success in unsupervised visual representation learning, which maximize the similarities between feature representations of different views of the same image, while minimize the similarities between feature representations of views of different images. In text summarization, the output summary is a shorter form of the input document and they have similar meanings. In this paper, we propose a contrastive learning model for supervised abstractive text summarization, where we view a document, its gold summary and its model generated summaries as different views of the same mean representation and maximize the similarities between them during training. We improve over a strong sequence-to-sequence text generation model (i.e., BART) on three different summarization datasets. Human evaluation also shows that our model achieves better faithfulness ratings compared to its counterpart without contrastive objectives.
Multimodal abstractive summarization (MAS) models that summarize videos (vision modality) and their corresponding transcripts (text modality) are able to extract the essential information from massive multimodal data on the Internet. Recently, large-scale generative pre-trained language models (GPLMs) have been shown to be effective in text generation tasks. However, existing MAS models cannot leverage GPLMs' powerful generation ability. To fill this research gap, we aim to study two research questions: 1) how to inject visual information into GPLMs without hurting their generation ability; and 2) where is the optimal place in GPLMs to inject the visual information? In this paper, we present a simple yet effective method to construct vision guided (VG) GPLMs for the MAS task using attention-based add-on layers to incorporate visual information while maintaining their original text generation ability. Results show that our best model significantly surpasses the prior state-of-the-art model by 5.7 ROUGE-1, 5.3 ROUGE-2, and 5.1 ROUGE-L scores on the How2 dataset, and our visual guidance method contributes 83.6% of the overall improvement. Furthermore, we conduct thorough ablation studies to analyze the effectiveness of various modality fusion methods and fusion locations.
Despite recent progress in abstractive summarization, systems still suffer from faithfulness errors. While prior work has proposed models that improve faithfulness, it is unclear whether the improvement comes from an increased level of extractiveness of the model outputs as one naive way to improve faithfulness is to make summarization models more extractive. In this work, we present a framework for evaluating the effective faithfulness of summarization systems, by generating a faithfulnessabstractiveness trade-off curve that serves as a control at different operating points on the abstractiveness spectrum. We then show that the Maximum Likelihood Estimation (MLE) baseline as well as a recently proposed method for improving faithfulness, are both worse than the control at the same level of abstractiveness. Finally, we learn a selector to identify the most faithful and abstractive summary for a given document, and show that this system can attain higher faithfulness scores in human evaluations while being more abstractive than the baseline system on two datasets. Moreover, we show that our system is able to achieve a better faithfulness-abstractiveness trade-off than the control at the same level of abstractiveness.
Despite significant progress has been achieved in text summarization, factual inconsistency in generated summaries still severely limits its practical applications. Among the key factors to ensure factual consistency, a reliable automatic evaluation metric is the first and the most crucial one. However, existing metrics either neglect the intrinsic cause of the factual inconsistency or rely on auxiliary tasks, leading to an unsatisfied correlation with human judgments or increasing the inconvenience of usage in practice. In light of these challenges, we propose a novel metric to evaluate the factual consistency in text summarization via counterfactual estimation, which formulates the causal relationship among the source document, the generated summary, and the language prior. We remove the effect of language prior, which can cause factual inconsistency, from the total causal effect on the generated summary, and provides a simple yet effective way to evaluate consistency without relying on other auxiliary tasks. We conduct a series of experiments on three public abstractive text summarization datasets, and demonstrate the advantages of the proposed metric in both improving the correlation with human judgments and the convenience of usage. The source code is available at https://github.com/xieyxclack/factual_coco.
In this paper, we present a denoising sequence-to-sequence (seq2seq) autoencoder via contrastive learning for abstractive text summarization. Our model adopts a standard Transformer-based architecture with a multi-layer bi-directional encoder and an auto-regressive decoder. To enhance its denoising ability, we incorporate self-supervised contrastive learning along with various sentence-level document augmentation. These two components, seq2seq autoencoder and contrastive learning, are jointly trained through fine-tuning, which improves the performance of text summarization with regard to ROUGE scores and human evaluation. We conduct experiments on two datasets and demonstrate that our model outperforms many existing benchmarks and even achieves comparable performance to the state-of-the-art abstractive systems trained with more complex architecture and extensive computation resources.
Encoder-decoder models have achieved remarkable success in abstractive text summarization, which aims to compress one or more documents into a shorter version without the loss of the essential content. Unfortunately, these models mostly suffer a discrepancy between training and inference, i.e., the exposure bias problem. During the training stage, with teacher forcing these models are optimized to maximize the likelihood of the gold summary given the gold summary tokens as input to the decoder, while at inference the given tokens are replaced by the generated tokens. Consequently, low-quality summaries are very likely to be generated. To remedy this problem, we propose to leverage contrastive learning to decrease the likelihood of these low-quality summaries, and meanwhile increase the likelihood of the gold summary. Since our solution expands the states that the model perceives during training, we expect that the exposure bias problem can be alleviated. We experimentally demonstrate that our method effectively improves the performance of the state-of-the-art model on different datasets.
Integrating multimodal knowledge for abstractive summarization task is a work-in-progress research area, with present techniques inheriting fusion-then-generation paradigm. Due to semantic gaps between computer vision and natural language processing, current methods often treat multiple data points as separate objects and rely on attention mechanisms to search for connection in order to fuse together. In addition, missing awareness of cross-modal matching from many frameworks leads to performance reduction. To solve these two drawbacks, we propose an Iterative Contrastive Alignment Framework (ICAF) that uses recurrent alignment and contrast to capture the coherences between images and texts. Specifically, we design a recurrent alignment (RA) layer to gradually investigate fine-grained semantical relationships between image patches and text tokens. At each step during the encoding process, cross-modal contrastive losses are applied to directly optimize the embedding space. According to ROUGE, relevance scores, and human evaluation, our model outperforms the state-of-the-art baselines on MSMO dataset. Experiments on the applicability of our proposed framework and hyperparameters settings have been also conducted.
Neural models for abstractive summarization tend to generate output that is fluent and well-formed but lacks semantic faithfulness, or factuality, with respect to the input documents. In this paper, we analyze the tradeoff between abstractiveness and factuality of generated summaries across multiple datasets and models, using extensive human evaluations of factuality. In our analysis, we visualize the rates of change in factuality as we gradually increase abstractiveness using a decoding constraint, and we observe that, while increased abstractiveness generally leads to a drop in factuality, the rate of factuality decay depends on factors such as the data that the system was trained on. We introduce two datasets with human factuality judgements; one containing 10.2k generated summaries with systematically varied degrees of abstractiveness; the other containing 4.2k summaries from five different summarization models. We propose new factuality metrics that adjust for the degree of abstractiveness, and we use them to compare the abstractiveness-adjusted factuality of previous summarization works, providing baselines for future work.