Modern summarization models generate highly fluent but often factually unreliable outputs. This motivated a surge of metrics attempting to measure the factuality of automatically generated summaries. Due to the lack of common benchmarks, these metrics cannot be compared. Moreover, all these methods treat factuality as a binary concept and fail to provide deeper insights into the kinds of inconsistencies made by different systems. To address these limitations, we devise a typology of factual errors and use it to collect human annotations of generated summaries from state-of-the-art summarization systems for the CNN/DM and XSum datasets. Through these annotations, we identify the proportion of different categories of factual errors in various summarization models and benchmark factuality metrics, showing their correlation with human judgment as well as their specific strengths and weaknesses.
Novel neural architectures, training strategies, and the availability of large-scale corpora haven been the driving force behind recent progress in abstractive text summarization. However, due to the black-box nature of neural models, uninformative evaluation metrics, and scarce tooling for model and data analysis, the true performance and failure modes of summarization models remain largely unknown. To address this limitation, we introduce SummVis, an open-source tool for visualizing abstractive summaries that enables fine-grained analysis of the models, data, and evaluation metrics associated with text summarization. Through its lexical and semantic visualizations, the tools offers an easy entry point for in-depth model prediction exploration across important dimensions such as factual consistency or abstractiveness. The tool together with several pre-computed model outputs is available at https://github.com/robustness-gym/summvis.
We introduce SummScreen, a summarization dataset comprised of pairs of TV series transcripts and human written recaps. The dataset provides a challenging testbed for abstractive summarization for several reasons. Plot details are often expressed indirectly in character dialogues and may be scattered across the entirety of the transcript. These details must be found and integrated to form the succinct plot descriptions in the recaps. Also, TV scripts contain content that does not directly pertain to the central plot but rather serves to develop characters or provide comic relief. This information is rarely contained in recaps. Since characters are fundamental to TV series, we also propose two entity-centric evaluation metrics. Empirically, we characterize the dataset by evaluating several methods, including neural models and those based on nearest neighbors. An oracle extractive approach outperforms all benchmarked models according to automatic metrics, showing that the neural models are unable to fully exploit the input transcripts. Human evaluation and qualitative analysis reveal that our non-oracle models are competitive with their oracle counterparts in terms of generating faithful plot events and can benefit from better content selectors. Both oracle and non-oracle models generate unfaithful facts, suggesting future research directions.
Recent pre-trained abstractive summarization systems have started to achieve credible performance, but a major barrier to their use in practice is their propensity to output summaries that are not faithful to the input and that contain factual errors. While a number of annotated datasets and statistical models for assessing factuality have been explored, there is no clear picture of what errors are most important to target or where current techniques are succeeding and failing. We explore both synthetic and human-labeled data sources for training models to identify factual errors in summarization, and study factuality at the word-, dependency-, and sentence-level. Our observations are threefold. First, exhibited factual errors differ significantly across datasets, and commonly-used training sets of simple synthetic errors do not reflect errors made on abstractive datasets like XSum. Second, human-labeled data with fine-grained annotations provides a more effective training signal than sentence-level annotations or synthetic data. Finally, we show that our best factuality detection model enables training of more factual XSum summarization models by allowing us to identify non-factual tokens in the training data.
State-of-the-art abstractive summarization models generally rely on extensive labeled data, which lowers their generalization ability on domains where such data are not available. In this paper, we present a study of domain adaptation for the abstractive summarization task across six diverse target domains in a low-resource setting. Specifically, we investigate the second phase of pre-training on large-scale generative models under three different settings: 1) source domain pre-training; 2) domain-adaptive pre-training; and 3) task-adaptive pre-training. Experiments show that the effectiveness of pre-training is correlated with the similarity between the pre-training data and the target domain task. Moreover, we find that continuing pre-training could lead to the pre-trained model's catastrophic forgetting, and a learning method with less forgetting can alleviate this issue. Furthermore, results illustrate that a huge gap still exists between the low-resource and high-resource settings, which highlights the need for more advanced domain adaptation methods for the abstractive summarization task.
There have been various types of pretraining architectures including autoencoding models (e.g., BERT), autoregressive models (e.g., GPT), and encoder-decoder models (e.g., T5). However, none of the pretraining frameworks performs the best for all tasks of three main categories including natural language understanding (NLU), unconditional generation, and conditional generation. We propose a General Language Model (GLM) based on autoregressive blank infilling to address this challenge. GLM improves blank filling pretraining by adding 2D positional encodings and allowing an arbitrary order to predict spans, which results in performance gains over BERT and T5 on NLU tasks. Meanwhile, GLM can be pretrained for different types of tasks by varying the number and lengths of blanks. On a wide range of tasks across NLU, conditional and unconditional generation, GLM outperforms BERT, T5, and GPT given the same model sizes and data, and achieves the best performance from a single pretrained model with 1.25x parameters of BERT Large , demonstrating its generalizability to different downstream tasks.
With sequentially stacked self-attention, (optional) encoder-decoder attention, and feed-forward layers, Transformer achieves big success in natural language processing (NLP), and many variants have been proposed. Currently, almost all these models assume that the layer order is fixed and kept the same across data samples. We observe that different data samples actually favor different orders of the layers. Based on this observation, in this work, we break the assumption of the fixed layer order in the Transformer and introduce instance-wise layer reordering into the model structure. Our Instance-wise Ordered Transformer (IOT) can model variant functions by reordered layers, which enables each sample to select the better one to improve the model performance under the constraint of almost the same number of parameters. To achieve this, we introduce a light predictor with negligible parameter and inference cost to decide the most capable and favorable layer order for any input sequence. Experiments on 3 tasks (neural machine translation, abstractive summarization, and code generation) and 9 datasets demonstrate consistent improvements of our method. We further show that our method can also be applied to other architectures beyond Transformer. Our code is released at Github.
A key challenge for abstractive summarization is ensuring factual consistency of the generated summary with respect to the original document. For example, state-of-the-art models trained on existing datasets exhibit entity hallucination, generating names of entities that are not present in the source document. We propose a set of new metrics to quantify the entity-level factual consistency of generated summaries and we show that the entity hallucination problem can be alleviated by simply filtering the training data. In addition, we propose a summary-worthy entity classification task to the training process as well as a joint entity and summary generation approach, which yield further improvements in entity level metrics.
Neural abstractive summarization has been studied in many pieces of literature and achieves great success with the aid of large corpora. However, when encountering novel tasks, one may not always benefit from transfer learning due to the domain shifting problem, and overfitting could happen without adequate labeled examples. Furthermore, the annotations of abstractive summarization are costly, which often demand domain knowledge to ensure the ground-truth quality. Thus, there are growing appeals for Low-Resource Abstractive Summarization, which aims to leverage past experience to improve the performance with limited labeled examples of target corpus. In this paper, we propose to utilize two knowledge-rich sources to tackle this problem, which are large pre-trained models and diverse existing corpora. The former can provide the primary ability to tackle summarization tasks; the latter can help discover common syntactic or semantic information to improve the generalization ability. We conduct extensive experiments on various summarization corpora with different writing styles and forms. The results demonstrate that our approach achieves the state-of-the-art on 6 corpora in low-resource scenarios, with only 0.7% of trainable parameters compared to previous work.
Abstractive summarization systems generally rely on large collections of document-summary pairs. However, the performance of abstractive systems remains a challenge due to the unavailability of parallel data for low-resource languages like Bengali. To overcome this problem, we propose a graph-based unsupervised abstractive summarization system in the single-document setting for Bengali text documents, which requires only a Part-Of-Speech (POS) tagger and a pre-trained language model trained on Bengali texts. We also provide a human-annotated dataset with document-summary pairs to evaluate our abstractive model and to support the comparison of future abstractive summarization systems of the Bengali Language. We conduct experiments on this dataset and compare our system with several well-established unsupervised extractive summarization systems. Our unsupervised abstractive summarization model outperforms the baselines without being exposed to any human-annotated reference summaries.
We propose pre-finetuning, an additional large-scale learning stage between language model pre-training and fine-tuning. Pre-finetuning is massively multi-task learning (around 50 datasets, over 4.8 million total labeled examples), and is designed to encourage learning of representations that generalize better to many different tasks. We show that pre-finetuning consistently improves performance for pretrained discriminators (e.g.~RoBERTa) and generation models (e.g.~BART) on a wide range of tasks (sentence prediction, commonsense reasoning, MRC, etc.), while also significantly improving sample efficiency during fine-tuning. We also show that large-scale multi-tasking is crucial; pre-finetuning can hurt performance when few tasks are used up until a critical point (usually above 15) after which performance improves linearly in the number of tasks.
Automatic text summarization has been widely studied as an important task in natural language processing. Traditionally, various feature engineering and machine learning based systems have been proposed for extractive as well as abstractive text summarization. Recently, deep learning based, specifically Transformer-based systems have been immensely popular. Summarization is a cognitively challenging task - extracting summary worthy sentences is laborious, and expressing semantics in brief when doing abstractive summarization is complicated. In this paper, we specifically look at the problem of summarizing scientific research papers from multiple domains. We differentiate between two types of summaries, namely, (a) LaySumm: A very short summary that captures the essence of the research paper in layman terms restricting overtly specific technical jargon and (b) LongSumm: A much longer detailed summary aimed at providing specific insights into various ideas touched upon in the paper. While leveraging latest Transformer-based models, our systems are simple, intuitive and based on how specific paper sections contribute to human summaries of the two types described above. Evaluations against gold standard summaries using ROUGE metrics prove the effectiveness of our approach. On blind test corpora, our system ranks first and third for the LongSumm and LaySumm tasks respectively.
Transformers have shown improved performance when compared to previous architectures for sequence processing such as RNNs. Despite their sizeable performance gains, as recently suggested, the model is computationally expensive to train and with a high parameter budget. In light of this, we explore parameter-sharing methods in Transformers with a specific focus on generative models. We perform an analysis of different parameter sharing/reduction methods and develop the Subformer. Our model combines sandwich-style parameter sharing, which overcomes naive cross-layer parameter sharing in generative models, and self-attentive embedding factorization (SAFE). Experiments on machine translation, abstractive summarization and language modeling show that the Subformer can outperform the Transformer even when using significantly fewer parameters.
Text summarization is one of the most critical Natural Language Processing (NLP) tasks. More and more researches are conducted in this field every day. Pre-trained transformer-based encoder-decoder models have begun to gain popularity for these tasks. This paper proposes two methods to address this task and introduces a novel dataset named pn-summary for Persian abstractive text summarization. The models employed in this paper are mT5 and an encoder-decoder version of the ParsBERT model (i.e., a monolingual BERT model for Persian). These models are fine-tuned on the pn-summary dataset. The current work is the first of its kind and, by achieving promising results, can serve as a baseline for any future work.
Information retrieval (IR) for precision medicine (PM) often involves looking for multiple pieces of evidence that characterize a patient case. This typically includes at least the name of a condition and a genetic variation that applies to the patient. Other factors such as demographic attributes, comorbidities, and social determinants may also be pertinent. As such, the retrieval problem is often formulated as ad hoc search but with multiple facets (e.g., disease, mutation) that may need to be incorporated. In this paper, we present a document reranking approach that combines neural query-document matching and text summarization toward such retrieval scenarios. Our architecture builds on the basic BERT model with three specific components for reranking: (a). document-query matching (b). keyword extraction and (c). facet-conditioned abstractive summarization. The outcomes of (b) and (c) are used to essentially transform a candidate document into a concise summary that can be compared with the query at hand to compute a relevance score. Component (a) directly generates a matching score of a candidate document for a query. The full architecture benefits from the complementary potential of document-query matching and the novel document transformation approach based on summarization along PM facets. Evaluations using NIST's TREC-PM track datasets (2017--2019) show that our model achieves state-of-the-art performance. To foster reproducibility, our code is made available here: https://github.com/bionlproc/text-summ-for-doc-retrieval.
The recent success of deep learning techniques for abstractive summarization is predicated on the availability of large-scale datasets. When summarizing reviews (e.g., for products or movies), such training data is neither available nor can be easily sourced, motivating the development of methods which rely on synthetic datasets for supervised training. We show that explicitly incorporating content planning in a summarization model not only yields output of higher quality, but also allows the creation of synthetic datasets which are more natural, resembling real world document-summary pairs. Our content plans take the form of aspect and sentiment distributions which we induce from data without access to expensive annotations. Synthetic datasets are created by sampling pseudo-reviews from a Dirichlet distribution parametrized by our content planner, while our model generates summaries based on input reviews and induced content plans. Experimental results on three domains show that our approach outperforms competitive models in generating informative, coherent, and fluent summaries that capture opinion consensus.
Abstractive summarization is the process of generating novel sentences based on the information extracted from the original text document while retaining the context. Due to abstractive summarization's underlying complexities, most of the past research work has been done on the extractive summarization approach. Nevertheless, with the triumph of the sequence-to-sequence (seq2seq) model, abstractive summarization becomes more viable. Although a significant number of notable research has been done in the English language based on abstractive summarization, only a couple of works have been done on Bengali abstractive news summarization (BANS). In this article, we presented a seq2seq based Long Short-Term Memory (LSTM) network model with attention at encoder-decoder. Our proposed system deploys a local attention-based model that produces a long sequence of words with lucid and human-like generated sentences with noteworthy information of the original document. We also prepared a dataset of more than 19k articles and corresponding human-written summaries collected from bangla.bdnews24.com1 which is till now the most extensive dataset for Bengali news document summarization and publicly published in Kaggle2. We evaluated our model qualitatively and quantitatively and compared it with other published results. It showed significant improvement in terms of human evaluation scores with state-of-the-art approaches for BANS.
Prior work in Argument Mining frequently alludes to its potential applications in automatic debating systems. Despite this focus, almost no datasets or models exist which apply natural language processing techniques to problems found within competitive formal debate. To remedy this, we present the DebateSum dataset. DebateSum consists of 187,386 unique pieces of evidence with corresponding argument and extractive summaries. DebateSum was made using data compiled by competitors within the National Speech and Debate Association over a 7-year period. We train several transformer summarization models to benchmark summarization performance on DebateSum. We also introduce a set of fasttext word-vectors trained on DebateSum called debate2vec. Finally, we present a search engine for this dataset which is utilized extensively by members of the National Speech and Debate Association today. The DebateSum search engine is available to the public here: http://www.debate.cards
Neural sequence models can generate highly fluent sentences, but recent studies have also shown that they are also prone to hallucinate additional content not supported by the input. These variety of fluent but wrong outputs are particularly problematic, as it will not be possible for users to tell they are being presented incorrect content. To detect these errors, we propose a task to predict whether each token in the output sequence is hallucinated (not contained in the input) and collect new manually annotated evaluation sets for this task. We also introduce a method for learning to detect hallucinations using pretrained language models fine tuned on synthetic data that includes automatically inserted hallucinations Experiments on machine translation (MT) and abstractive summarization demonstrate that our proposed approach consistently outperforms strong baselines on all benchmark datasets. We further demonstrate how to use the token-level hallucination labels to define a fine-grained loss over the target sequence in low-resource MT and achieve significant improvements over strong baseline methods. We also apply our method to word-level quality estimation for MT and show its effectiveness in both supervised and unsupervised settings. Codes and data available at https://github.com/violet-zct/fairseq-detect-hallucination.
In the Query Focused Multi-Document Summarization (QF-MDS) task, a set of documents and a query are given where the goal is to generate a summary from these documents based on the given query. However, one major challenge for this task is the lack of availability of labeled training datasets. To overcome this issue, in this paper, we propose a novel weakly supervised learning approach via utilizing distant supervision. In particular, we use datasets similar to the target dataset as the training data where we leverage pre-trained sentence similarity models to generate the weak reference summary of each individual document in a document set from the multi-document gold reference summaries. Then, we iteratively train our summarization model on each single-document to alleviate the computational complexity issue that occurs while training neural summarization models in multiple documents (i.e., long sequences) at once. Experimental results in Document Understanding Conferences (DUC) datasets show that our proposed approach sets a new state-of-the-art result in terms of various evaluation metrics.