The importance of building semantic parsers which can be applied to new domains and generate programs unseen at training has long been acknowledged, and datasets testing out-of-domain performance are becoming increasingly available. However, little or no attention has been devoted to studying learning algorithms or objectives which promote domain generalization, with virtually all existing approaches relying on standard supervised learning. In this work, we use a meta-learning framework which targets specifically zero-shot domain generalization for semantic parsing. We apply a model-agnostic training algorithm that simulates zero-shot parsing by constructing virtual train and test sets from disjoint domains. The learning objective capitalizes on the intuition that gradient steps that improve source-domain performance should also improve target-domain performance, thus encouraging a parser to generalize well to unseen target domains. Experimental results on the (English) Spider and Chinese Spider datasets show that the meta-learning objective significantly boosts the performance of a baseline parser.
Although neural sequence-to-sequence models have been successfully applied to semantic parsing, they struggle to perform well on query-based data splits that require \emph{composition generalization}, an ability of systematically generalizing to unseen composition of seen components. Motivated by the explicitly built-in compositionality in traditional statistical semantic parsing, we propose a new decoding framework that preserves the expressivity and generality of sequence-to-sequence models while featuring explicit lexicon-style alignments and disentangled information processing. Specifically, we decompose decoding into two phases where an input utterance is first tagged with semantic symbols representing the meanings of its individual words, and then a sequence-to-sequence model is used to predict the final meaning representation conditioning on the utterance and the predicted tag sequence. Experimental results on three semantic parsing datasets with query-based splits show that the proposed approach consistently improves compositional generalization of sequence-to-sequence models across different model architectures, domains and semantic formalisms.
In this paper we apply self-knowledge distillation to text summarization which we argue can alleviate problems with maximum-likelihood training on single reference and noisy datasets. Instead of relying on one-hot annotation labels, our student summarization model is trained with guidance from a teacher which generates smoothed labels to help regularize training. Furthermore, to better model uncertainty during training, we introduce multiple noise signals for both teacher and student models. We demonstrate experimentally on three benchmarks that our framework boosts the performance of both pretrained and non-pretrained summarizers achieving state-of-the-art results.
Opinion summarization is an automatic creation of text reflecting subjective information expressed in multiple documents, such as user reviews of a product. The task is practically important and has attracted a lot of attention. However, due to a high cost of summary production, datasets large enough for training supervised models are lacking. Instead, the task has been traditionally approached with extractive methods that learn to select text fragments in an unsupervised or weakly-supervised way. Recently, it has been shown that abstractive summaries, potentially more fluent and better at reflecting conflicting information, can also be produced in an unsupervised fashion. However, these models, not being exposed to the actual summaries, fail to capture their essential properties. In this work, we show that even a handful of summaries is sufficient to bootstrap generation of the summary text with all expected properties, such as writing style, informativeness, fluency, and sentiment preservation. We start by training a language model to generate a new product review given available reviews of the product. The model is aware of the properties: it proceeds with first generating property values and then producing a review conditioned on them. We do not use any summaries in this stage and the property values are derived from reviews with no manual effort. In the second stage, we fine-tune the module predicting the property values on a few available summaries. This lets us switch the generator to the summarization mode. Our approach substantially outperforms previous extractive and abstractive methods in automatic and human evaluation.
Most general-purpose extractive summarization models are trained on news articles, which are short and present all important information upfront. As a result, such models are biased on position and often perform a smart selection of sentences from the beginning of the document. When summarizing long narratives, which have complex structure and present information piecemeal, simple position heuristics are not sufficient. In this paper, we propose to explicitly incorporate the underlying structure of narratives into general unsupervised and supervised extractive summarization models. We formalize narrative structure in terms of key narrative events (turning points) and treat it as latent in order to summarize screenplays (i.e., extract an optimal sequence of scenes). Experimental results on the CSI corpus of TV screenplays, which we augment with scene-level summarization labels, show that latent turning points correlate with important aspects of a CSI episode and improve summarization performance over general extractive algorithms leading to more complete and diverse summaries.
It is a big challenge to model long-range input for document summarization. In this paper, we target using a select and generate paradigm to enhance the capability of selecting explainable contents (i.e., interpret the selection given its semantics, novelty, relevance) and then guiding to control the abstract generation. Specifically, a newly designed pair-wise extractor is proposed to capture the sentence pair interactions and their centrality. Furthermore, the generator is hybrid with the selected content and is jointly integrated with a pointer distribution that is derived from a sentence deployment's attention. The abstract generation can be controlled by an explainable mask matrix that determines to what extent the content can be included in the summary. Encoders are adaptable with both Transformer-based and BERT-based configurations. Overall, both results based on ROUGE metrics and human evaluation gain outperformance over several state-of-the-art models on two benchmark CNN/DailyMail and NYT datasets.
Successful linguistic communication relies on a shared experience of the world, and it is this shared experience that makes utterances meaningful. Despite the incredible effectiveness of language processing models trained on text alone, today's best systems still make mistakes that arise from a failure to relate language to the physical world it describes and to the social interactions it facilitates. Natural Language Processing is a diverse field, and progress throughout its development has come from new representational theories, modeling techniques, data collection paradigms, and tasks. We posit that the present success of representation learning approaches trained on large text corpora can be deeply enriched from the parallel tradition of research on the contextual and social nature of language. In this article, we consider work on the contextual foundations of language: grounding, embodiment, and social interaction. We describe a brief history and possible progression of how contextual information can factor into our representations, with an eye towards how this integration can move the field forward and where it is currently being pioneered. We believe this framing will serve as a roadmap for truly contextual language understanding.
The supervised training of high-capacity models on large datasets containing hundreds of thousands of document-summary pairs is critical to the recent success of deep learning techniques for abstractive summarization. Unfortunately, in most domains (other than news) such training data is not available and cannot be easily sourced. In this paper we enable the use of supervised learning for the setting where there are only documents available (e.g.,~product or business reviews) without ground truth summaries. We create a synthetic dataset from a corpus of user reviews by sampling a review, pretending it is a summary, and generating noisy versions thereof which we treat as pseudo-review input. We introduce several linguistically motivated noise generation functions and a summarization model which learns to denoise the input and generate the original review. At test time, the model accepts genuine reviews and generates a summary containing salient opinions, treating those that do not reach consensus as noise. Extensive automatic and human evaluation shows that our model brings substantial improvements over both abstractive and extractive baselines.
Datasets for semantic parsing scarcely consider languages other than English and professional translation can be prohibitively expensive. In this work, we propose to adapt a semantic parser trained on a single language, such as English, to new languages and multiple domains with minimal annotation. We evaluate if machine translation is an adequate substitute for training data, and extend this to investigate bootstrapping using joint training with English, paraphrasing, and resources such as multilingual BERT. Experimental results on a new version of ATIS and Overnight in German and Chinese indicate that MT can approximate training data in a new language for accurate parsing when augmented with paraphrasing through multiple MT engines.