NAACL 2021 Papers

We present a method for constructing taxonomic trees (e.g., WordNet) using pretrained language models. Our approach is composed of two modules, one that predicts parenthood relations and another that reconciles those pairwise predictions into trees. The parenthood prediction module produces likelihood scores for each potential parent-child pair, creating a graph of parent-child relation scores. The tree reconciliation module treats the task as a graph optimization problem and outputs the maximum spanning tree of this graph. We train our model on subtrees sampled from WordNet, and test on nonoverlapping WordNet subtrees. We show that incorporating web-retrieved glosses can further improve performance. On the task of constructing subtrees of English WordNet, the model achieves 66.7 ancestor F1, a 20.0% relative increase over the previous best published result on this task. In addition, we convert the original English dataset into nine other languages using Open Multilingual WordNet and extend our results across these languages.

We propose a neural event coreference model in which event coreference is jointly trained with five tasks: trigger detection, entity coreference, anaphoricity determination, realis detection, and argument extraction. To guide the learning of this complex model, we incorporate cross-task consistency constraints into the learning process as soft constraints via designing penalty functions. In addition, we propose the novel idea of viewing entity coreference and event coreference as a single coreference task, which we believe is a step towards a unified model of coreference resolution. The resulting model achieves state-of-the-art results on the KBP 2017 event coreference dataset.

Adversarial attacks alter NLP model predictions by perturbing test-time inputs. However, it is much less understood whether, and how, predictions can be manipulated with small, concealed changes to the training data. In this work, we develop a new data poisoning attack that allows an adversary to control model predictions whenever a desired trigger phrase is present in the input. For instance, we insert 50 poison examples into a sentiment model’s training set that causes the model to frequently predict Positive whenever the input contains “James Bond”. Crucially, we craft these poison examples using a gradient-based procedure so that they do not mention the trigger phrase. We also apply our poison attack to language modeling (“Apple iPhone” triggers negative generations) and machine translation (“iced coffee” mistranslated as “hot coffee”). We conclude by proposing three defenses that can mitigate our attack at some cost in prediction accuracy or extra human annotation.

We propose ConVEx (Conversational Value Extractor), an efficient pretraining and fine-tuning neural approach for slot-labeling dialog tasks. Instead of relying on more general pretraining objectives from prior work (e.g., language modeling, response selection), ConVEx’s pretraining objective, a novel pairwise cloze task using Reddit data, is well aligned with its intended usage on sequence labeling tasks. This enables learning domain-specific slot labelers by simply fine-tuning decoding layers of the pretrained general-purpose sequence labeling model, while the majority of the pretrained model’s parameters are kept frozen. We report state-of-the-art performance of ConVEx across a range of diverse domains and data sets for dialog slot-labeling, with the largest gains in the most challenging, few-shot setups. We believe that ConVEx’s reduced pretraining times (i.e., only 18 hours on 12 GPUs) and cost, along with its efficient fine-tuning and strong performance, promise wider portability and scalability for data-efficient sequence-labeling tasks in general.

We describe a span-level supervised attention loss that improves compositional generalization in semantic parsers. Our approach builds on existing losses that encourage attention maps in neural sequence-to-sequence models to imitate the output of classical word alignment algorithms. Where past work has used word-level alignments, we focus on spans; borrowing ideas from phrase-based machine translation, we align subtrees in semantic parses to spans of input sentences, and encourage neural attention mechanisms to mimic these alignments. This method improves the performance of transformers, RNNs, and structured decoders on three benchmarks of compositional generalization.

Recent advances in transfer learning have improved the performance of virtual assistants considerably. Nevertheless, creating sophisticated voice-enabled applications for new domains remains a challenge, and meager training data is often a key bottleneck. Accordingly, unsupervised learning and SSL (semi-supervised learning) techniques continue to be of vital importance. While a number of such methods have been explored previously in isolation, in this paper we investigate the synergistic use of a number of weakly supervised techniques with a view to improving NLU (Natural Language Understanding) accuracy in low-resource settings. We explore three different approaches incorporating anonymized, unlabeled and automatically transcribed user utterances into the training process, two focused on data augmentation via SSL and another one focused on unsupervised and transfer learning. We show promising results, obtaining gains that range from 4.73% to 7.65% relative improvements on semantic error rate for each individual approach. Moreover, the combination of all three methods together yields a relative improvement of 11.77% over our current baseline model. Our methods are applicable to any new domain with minimal training data, and can be deployed over time into a cycle of continual learning.

Neural models for text generation are often designed in an end-to-end fashion, typically with zero control over intermediate computations, limiting their practical usability in downstream applications. In this work, we incorporate explicit means into neural models to ensure topical continuity, informativeness and content diversity of generated radiology reports. For the purpose we propose a method to compute image representations specific to each sentential context and eliminate redundant content by exploiting diverse sentence states. We conduct experiments to generate radiology reports from medical images of chest x-rays using MIMIC-CXR. Our model outperforms baselines by up to 18% and 29% respective in the evaluation for informativeness and content ordering respectively, relative on objective metrics and 16% on human evaluation.

Multilingual models have demonstrated impressive cross-lingual transfer performance. However, test sets like XNLI are monolingual at the example level. In multilingual communities, it is common for polyglots to code-mix when conversing with each other. Inspired by this phenomenon, we present two strong black-box adversarial attacks (one word-level, one phrase-level) for multilingual models that push their ability to handle code-mixed sentences to the limit. The former uses bilingual dictionaries to propose perturbations and translations of the clean example for sense disambiguation. The latter directly aligns the clean example with its translations before extracting phrases as perturbations. Our phrase-level attack has a success rate of 89.75% against XLM-R-large, bringing its average accuracy of 79.85 down to 8.18 on XNLI. Finally, we propose an efficient adversarial training scheme that trains in the same number of steps as the original model and show that it creates more language-invariant representations, improving clean and robust accuracy in the absence of lexical overlap without degrading performance on the original examples.

Many recent approaches towards neural information retrieval mitigate their computational costs by using a multi-stage ranking pipeline. In the first stage, a number of potentially relevant candidates are retrieved using an efficient retrieval model such as BM25. Although BM25 has proven decent performance as a first-stage ranker, it tends to miss relevant passages. In this context we propose CoRT, a simple neural first-stage ranking model that leverages contextual representations from pretrained language models such as BERT to complement term-based ranking functions while causing no significant delay at query time. Using the MS MARCO dataset, we show that CoRT significantly increases the candidate recall by complementing BM25 with missing candidates. Consequently, we find subsequent re-rankers achieve superior results with less candidates. We further demonstrate that passage retrieval using CoRT can be realized with surprisingly low latencies.

Due to large number of entities in biomedical knowledge bases, only a small fraction of entities have corresponding labelled training data. This necessitates entity linking models which are able to link mentions of unseen entities using learned representations of entities. Previous approaches link each mention independently, ignoring the relationships within and across documents between the entity mentions. These relations can be very useful for linking mentions in biomedical text where linking decisions are often difficult due mentions having a generic or a highly specialized form. In this paper, we introduce a model in which linking decisions can be made not merely by linking to a knowledge base entity but also by grouping multiple mentions together via clustering and jointly making linking predictions. In experiments on the largest publicly available biomedical dataset, we improve the best independent prediction for entity linking by 3.0 points of accuracy, and our clustering-based inference model further improves entity linking by 2.3 points.

Training dialogue agents requires a large number of interactions with users: agents have no idea about which responses are bad among a lengthy dialogue. In this paper, we propose loop-clipping policy optimisation (LCPO) to eliminate useless responses. LCPO consists of two stages: loop clipping and advantage clipping. In loop clipping, we clip off useless responses (called loops) from dialogue history (called trajectories). The clipped trajectories are more succinct than the original ones, and the estimation of state-value is more accurate. Second, in advantage clipping, we estimate and clip the advantages of useless responses and normal ones separately. The clipped advantage distinguish useless actions from others and reduce the probabilities of useless actions efficiently. In experiments on Cambridge Restaurant Dialogue System, LCPO uses only 260 training dialogues to achieve 80% success rate, while PPO baseline requires 2160 dialogues. Besides, LCPO receives 3.7/5 scores in human evaluation where the agent interactively collects 100 real-user dialogues in training phase.

Story generation is an open-ended and subjective task, which poses a challenge for evaluating story generation models. We present Choose Your Own Adventure, a collaborative writing setup for pairwise model evaluation. Two models generate suggestions to people as they write a short story; we ask writers to choose one of the two suggestions, and we observe which model’s suggestions they prefer. The setup also allows further analysis based on the revisions people make to the suggestions. We show that these measures, combined with automatic metrics, provide an informative picture of the models’ performance, both in cases where the differences in generation methods are small (nucleus vs. top-k sampling) and large (GPT2 vs. Fusion models).

Computational linguistic research on language change through distributional semantic (DS) models has inspired researchers from fields such as philosophy and literary studies, who use these methods for the exploration and comparison of comparatively small datasets traditionally analyzed by close reading. Research on methods for small data is still in early stages and it is not clear which methods achieve the best results. We investigate the possibilities and limitations of using distributional semantic models for analyzing philosophical data by means of a realistic use-case. We provide a ground truth for evaluation created by philosophy experts and a blueprint for using DS models in a sound methodological setup. We compare three methods for creating specialized models from small datasets. Though the models do not perform well enough to directly support philosophers yet, we find that models designed for small data yield promising directions for future work.

The robustness and security of natural language processing (NLP) models are significantly important in real-world applications. In the context of text classification tasks, adversarial examples can be designed by substituting words with synonyms under certain semantic and syntactic constraints, such that a well-trained model will give a wrong prediction. Therefore, it is crucial to develop techniques to provide a rigorous and provable robustness guarantee against such attacks. In this paper, we propose WordDP to achieve certified robustness against word substitution at- tacks in text classification via differential privacy (DP). We establish the connection between DP and adversarial robustness for the first time in the text domain and propose a conceptual exponential mechanism-based algorithm to formally achieve the robustness. We further present a practical simulated exponential mechanism that has efficient inference with certified robustness. We not only provide a rigorous analytic derivation of the certified condition but also experimentally compare the utility of WordDP with existing defense algorithms. The results show that WordDP achieves higher accuracy and more than 30X efficiency improvement over the state-of-the-art certified robustness mechanism in typical text classification tasks.

We consider the problem of using observational data to estimate the causal effects of linguistic properties. For example, does writing a complaint politely lead to a faster response time? How much will a positive product review increase sales? This paper addresses two technical challenges related to the problem before developing a practical method. First, we formalize the causal quantity of interest as the effect of a writer’s intent, and establish the assumptions necessary to identify this from observational data. Second, in practice, we only have access to noisy proxies for the linguistic properties of interest—e.g., predictions from classifiers and lexicons. We propose an estimator for this setting and prove that its bias is bounded when we perform an adjustment for the text. Based on these results, we introduce TextCause, an algorithm for estimating causal effects of linguistic properties. The method leverages (1) distant supervision to improve the quality of noisy proxies, and (2) a pre-trained language model (BERT) to adjust for the text. We show that the proposed method outperforms related approaches when estimating the effect of Amazon review sentiment on semi-simulated sales figures. Finally, we present an applied case study investigating the effects of complaint politeness on bureaucratic response times.

Recent work in natural language processing (NLP) has focused on ethical challenges such as understanding and mitigating bias in data and algorithms; identifying objectionable content like hate speech, stereotypes and offensive language; and building frameworks for better system design and data handling practices. However, there has been little discussion about the ethical foundations that underlie these efforts. In this work, we study one ethical theory, namely deontological ethics, from the perspective of NLP. In particular, we focus on the generalization principle and the respect for autonomy through informed consent. We provide four case studies to demonstrate how these principles can be used with NLP systems. We also recommend directions to avoid the ethical issues in these systems.

Transformer based architectures are recently used for the task of answering questions over tables. In order to improve the accuracy on this task, specialized pre-training techniques have been developed and applied on millions of open-domain web tables. In this paper, we propose two novel approaches demonstrating that one can achieve superior performance on table QA task without even using any of these specialized pre-training techniques. The first model, called RCI interaction, leverages a transformer based architecture that independently classifies rows and columns to identify relevant cells. While this model yields extremely high accuracy at finding cell values on recent benchmarks, a second model we propose, called RCI representation, provides a significant efficiency advantage for online QA systems over tables by materializing embeddings for existing tables. Experiments on recent benchmarks prove that the proposed methods can effectively locate cell values on tables (up to ~98% Hit@1 accuracy on WikiSQL lookup questions). Also, the interaction model outperforms the state-of-the-art transformer based approaches, pre-trained on very large table corpora (TAPAS and TaBERT), achieving ~3.4% and ~18.86% additional precision improvement on the standard WikiSQL benchmark.

Latent alignment objectives such as CTC and AXE significantly improve non-autoregressive machine translation models. Can they improve autoregressive models as well? We explore the possibility of training autoregressive machine translation models with latent alignment objectives, and observe that, in practice, this approach results in degenerate models. We provide a theoretical explanation for these empirical results, and prove that latent alignment objectives are incompatible with teacher forcing.

Automated systems that negotiate with humans have broad applications in pedagogy and conversational AI. To advance the development of practical negotiation systems, we present CaSiNo: a novel corpus of over a thousand negotiation dialogues in English. Participants take the role of campsite neighbors and negotiate for food, water, and firewood packages for their upcoming trip. Our design results in diverse and linguistically rich negotiations while maintaining a tractable, closed-domain environment. Inspired by the literature in human-human negotiations, we annotate persuasion strategies and perform correlation analysis to understand how the dialogue behaviors are associated with the negotiation performance. We further propose and evaluate a multi-task framework to recognize these strategies in a given utterance. We find that multi-task learning substantially improves the performance for all strategy labels, especially for the ones that are the most skewed. We release the dataset, annotations, and the code to propel future work in human-machine negotiations: https://github.com/kushalchawla/CaSiNo