ACL 2020 Papers

Understanding human preferences, along with cultural and social nuances, lives at the heart of natural language understanding. Concretely, we present a new task and corpus for learning alignments between machine and human preferences. Our newly introduced problem is concerned with predicting the preferable options from two sentences describing scenarios that may involve social and cultural situations. Our problem is framed as a natural language inference task with crowd-sourced preference votes by human players, obtained from a gamified voting platform. We benchmark several state-of-the-art neural models, along with BERT and friends on this task. Our experimental results show that current state-of-the-art NLP models still leave much room for improvement.

While national politics often receive the spotlight, the overwhelming majority of legislation proposed, discussed, and enacted is done at the state level. Despite this fact, there is little awareness of the dynamics that lead to adopting these policies. In this paper, we take the first step towards a better understanding of these processes and the underlying dynamics that shape them, using data-driven methods. We build a new large-scale dataset, from multiple data sources, connecting state bills and legislator information, geographical information about their districts, and donations and donors’ information. We suggest a novel task, predicting the legislative body’s vote breakdown for a given bill, according to different criteria of interest, such as gender, rural-urban and ideological splits. Finally, we suggest a shared relational embedding model, representing the interactions between the text of the bill and the legislative context in which it is presented. Our experiments show that providing this context helps improve the prediction over strong text-based models.

Ideal point models analyze lawmakers’ votes to quantify their political positions, or ideal points. But votes are not the only way to express a political position. Lawmakers also give speeches, release press statements, and post tweets. In this paper, we introduce the text-based ideal point model (TBIP), an unsupervised probabilistic topic model that analyzes texts to quantify the political positions of its authors. We demonstrate the TBIP with two types of politicized text data: U.S. Senate speeches and senator tweets. Though the model does not analyze their votes or political affiliations, the TBIP separates lawmakers by party, learns interpretable politicized topics, and infers ideal points close to the classical vote-based ideal points. One benefit of analyzing texts, as opposed to votes, is that the TBIP can estimate ideal points of anyone who authors political texts, including non-voting actors. To this end, we use it to study tweets from the 2020 Democratic presidential candidates. Using only the texts of their tweets, it identifies them along an interpretable progressive-to-moderate spectrum.

Many applications of computational social science aim to infer causal conclusions from non-experimental data. Such observational data often contains confounders, variables that influence both potential causes and potential effects. Unmeasured or latent confounders can bias causal estimates, and this has motivated interest in measuring potential confounders from observed text. For example, an individual’s entire history of social media posts or the content of a news article could provide a rich measurement of multiple confounders. Yet, methods and applications for this problem are scattered across different communities and evaluation practices are inconsistent. This review is the first to gather and categorize these examples and provide a guide to data-processing and evaluation decisions. Despite increased attention on adjusting for confounding using text, there are still many open problems, which we highlight in this paper.

People vary in their ability to make accurate predictions about the future. Prior studies have shown that some individuals can predict the outcome of future events with consistently better accuracy. This leads to a natural question: what makes some forecasters better than others? In this paper we explore connections between the language people use to describe their predictions and their forecasting skill. Datasets from two different forecasting domains are explored: (1) geopolitical forecasts from Good Judgment Open, an online prediction forum and (2) a corpus of company earnings forecasts made by financial analysts. We present a number of linguistic metrics which are computed over text associated with people’s predictions about the future including: uncertainty, readability, and emotion. By studying linguistic factors associated with predictions, we are able to shed some light on the approach taken by skilled forecasters. Furthermore, we demonstrate that it is possible to accurately predict forecasting skill using a model that is based solely on language. This could potentially be useful for identifying accurate predictions or potentially skilled forecasters earlier.

Not all documents are equally important. Language processing is increasingly finding use as a supplement for questionnaires to assess psychological attributes of consenting individuals, but most approaches neglect to consider whether all documents of an individual are equally informative. In this paper, we present a novel model that uses message-level attention to learn the relative weight of users’ social media posts for assessing their five factor personality traits. We demonstrate that models with message-level attention outperform those with word-level attention, and ultimately yield state-of-the-art accuracies for all five traits by using both word and message attention in combination with past approaches (an average increase in Pearson r of 2.5%). In addition, examination of the high-signal posts identified by our model provides insight into the relationship between language and personality, helping to inform future work.

Natural disasters (e.g., hurricanes) affect millions of people each year, causing widespread destruction in their wake. People have recently taken to social media websites (e.g., Twitter) to share their sentiments and feelings with the larger community. Consequently, these platforms have become instrumental in understanding and perceiving emotions at scale. In this paper, we introduce HurricaneEmo, an emotion dataset of 15,000 English tweets spanning three hurricanes: Harvey, Irma, and Maria. We present a comprehensive study of fine-grained emotions and propose classification tasks to discriminate between coarse-grained emotion groups. Our best BERT model, even after task-guided pre-training which leverages unlabeled Twitter data, achieves only 68% accuracy (averaged across all groups). HurricaneEmo serves not only as a challenging benchmark for models but also as a valuable resource for analyzing emotions in disaster-centric domains.

Throughout a conversation, participants make choices that can orient the flow of the interaction. Such choices are particularly salient in the consequential domain of crisis counseling, where a difficulty for counselors is balancing between two key objectives: advancing the conversation towards a resolution, and empathetically addressing the crisis situation. In this work, we develop an unsupervised methodology to quantify how counselors manage this balance. Our main intuition is that if an utterance can only receive a narrow range of appropriate replies, then its likely aim is to advance the conversation forwards, towards a target within that range. Likewise, an utterance that can only appropriately follow a narrow range of possible utterances is likely aimed backwards at addressing a specific situation within that range. By applying this intuition, we can map each utterance to a continuous orientation axis that captures the degree to which it is intended to direct the flow of the conversation forwards or backwards. This unsupervised method allows us to characterize counselor behaviors in a large dataset of crisis counseling conversations, where we show that known counseling strategies intuitively align with this axis. We also illustrate how our measure can be indicative of a conversation’s progress, as well as its effectiveness.

Pre-trained visually grounded language models such as ViLBERT, LXMERT, and UNITER have achieved significant performance improvement on vision-and-language tasks but what they learn during pre-training remains unclear. In this work, we demonstrate that certain attention heads of a visually grounded language model actively ground elements of language to image regions. Specifically, some heads can map entities to image regions, performing the task known as entity grounding. Some heads can even detect the syntactic relations between non-entity words and image regions, tracking, for example, associations between verbs and regions corresponding to their arguments. We denote this ability as syntactic grounding. We verify grounding both quantitatively and qualitatively, using Flickr30K Entities as a testbed.

An increasing number of natural language processing papers address the effect of bias on predictions, introducing mitigation techniques at different parts of the standard NLP pipeline (data and models). However, these works have been conducted individually, without a unifying framework to organize efforts within the field. This situation leads to repetitive approaches, and focuses overly on bias symptoms/effects, rather than on their origins, which could limit the development of effective countermeasures. In this paper, we propose a unifying predictive bias framework for NLP. We summarize the NLP literature and suggest general mathematical definitions of predictive bias. We differentiate two consequences of bias: outcome disparities and error disparities, as well as four potential origins of biases: label bias, selection bias, model overamplification, and semantic bias. Our framework serves as an overview of predictive bias in NLP, integrating existing work into a single structure, and providing a conceptual baseline for improved frameworks.

While pretrained models such as BERT have shown large gains across natural language understanding tasks, their performance can be improved by further training the model on a data-rich intermediate task, before fine-tuning it on a target task. However, it is still poorly understood when and why intermediate-task training is beneficial for a given target task. To investigate this, we perform a large-scale study on the pretrained RoBERTa model with 110 intermediate-target task combinations. We further evaluate all trained models with 25 probing tasks meant to reveal the specific skills that drive transfer. We observe that intermediate tasks requiring high-level inference and reasoning abilities tend to work best. We also observe that target task performance is strongly correlated with higher-level abilities such as coreference resolution. However, we fail to observe more granular correlations between probing and target task performance, highlighting the need for further work on broad-coverage probing benchmarks. We also observe evidence that the forgetting of knowledge learned during pretraining may limit our analysis, highlighting the need for further work on transfer learning methods in these settings.

Legal Artificial Intelligence (LegalAI) focuses on applying the technology of artificial intelligence, especially natural language processing, to benefit tasks in the legal domain. In recent years, LegalAI has drawn increasing attention rapidly from both AI researchers and legal professionals, as LegalAI is beneficial to the legal system for liberating legal professionals from a maze of paperwork. Legal professionals often think about how to solve tasks from rule-based and symbol-based methods, while NLP researchers concentrate more on data-driven and embedding methods. In this paper, we introduce the history, the current state, and the future directions of research in LegalAI. We illustrate the tasks from the perspectives of legal professionals and NLP researchers and show several representative applications in LegalAI. We conduct experiments and provide an in-depth analysis of the advantages and disadvantages of existing works to explore possible future directions. You can find the implementation of our work from https://github.com/thunlp/CLAIM.

This position paper describes and critiques the Pretraining-Agnostic Identically Distributed (PAID) evaluation paradigm, which has become a central tool for measuring progress in natural language understanding. This paradigm consists of three stages: (1) pre-training of a word prediction model on a corpus of arbitrary size; (2) fine-tuning (transfer learning) on a training set representing a classification task; (3) evaluation on a test set drawn from the same distribution as that training set. This paradigm favors simple, low-bias architectures, which, first, can be scaled to process vast amounts of data, and second, can capture the fine-grained statistical properties of a particular data set, regardless of whether those properties are likely to generalize to examples of the task outside the data set. This contrasts with humans, who learn language from several orders of magnitude less data than the systems favored by this evaluation paradigm, and generalize to new tasks in a consistent way. We advocate for supplementing or replacing PAID with paradigms that reward architectures that generalize as quickly and robustly as humans.

We extracted information from the ACL Anthology (AA) and Google Scholar (GS) to examine trends in citations of NLP papers. We explore questions such as: how well cited are papers of different types (journal articles, conference papers, demo papers, etc.)? how well cited are papers from different areas of within NLP? etc. Notably, we show that only about 56% of the papers in AA are cited ten or more times. CL Journal has the most cited papers, but its citation dominance has lessened in recent years. On average, long papers get almost three times as many citations as short papers; and papers on sentiment classification, anaphora resolution, and entity recognition have the highest median citations. The analyses presented here, and the associated dataset of NLP papers mapped to citations, have a number of uses including: understanding how the field is growing and quantifying the impact of different types of papers.

The success of the large neural language models on many NLP tasks is exciting. However, we find that these successes sometimes lead to hype in which these models are being described as “understanding” language or capturing “meaning”. In this position paper, we argue that a system trained only on form has a priori no way to learn meaning. In keeping with the ACL 2020 theme of “Taking Stock of Where We’ve Been and Where We’re Going”, we argue that a clear understanding of the distinction between form and meaning will help guide the field towards better science around natural language understanding.

Human speakers have an extensive toolkit of ways to express themselves. In this paper, we engage with an idea largely absent from discussions of meaning in natural language understanding—namely, that the way something is expressed reflects different ways of conceptualizing or construing the information being conveyed. We first define this phenomenon more precisely, drawing on considerable prior work in theoretical cognitive semantics and psycholinguistics. We then survey some dimensions of construed meaning and show how insights from construal could inform theoretical and practical work in NLP.

Opinion summarization is the task of automatically creating summaries that reflect subjective information expressed in multiple documents, such as product reviews. While the majority of previous work has focused on the extractive setting, i.e., selecting fragments from input reviews to produce a summary, we let the model generate novel sentences and hence produce abstractive summaries. Recent progress in summarization has seen the development of supervised models which rely on large quantities of document-summary pairs. Since such training data is expensive to acquire, we instead consider the unsupervised setting, in other words, we do not use any summaries in training. We define a generative model for a review collection which capitalizes on the intuition that when generating a new review given a set of other reviews of a product, we should be able to control the “amount of novelty” going into the new review or, equivalently, vary the extent to which it deviates from the input. At test time, when generating summaries, we force the novelty to be minimal, and produce a text reflecting consensus opinions. We capture this intuition by defining a hierarchical variational autoencoder model. Both individual reviews and the products they correspond to are associated with stochastic latent codes, and the review generator (“decoder”) has direct access to the text of input reviews through the pointer-generator mechanism. Experiments on Amazon and Yelp datasets, show that setting at test time the review’s latent code to its mean, allows the model to produce fluent and coherent summaries reflecting common opinions.

This work presents a new approach to unsupervised abstractive summarization based on maximizing a combination of coverage and fluency for a given length constraint. It introduces a novel method that encourages the inclusion of key terms from the original document into the summary: key terms are masked out of the original document and must be filled in by a coverage model using the current generated summary. A novel unsupervised training procedure leverages this coverage model along with a fluency model to generate and score summaries. When tested on popular news summarization datasets, the method outperforms previous unsupervised methods by more than 2 R-1 points, and approaches results of competitive supervised methods. Our model attains higher levels of abstraction with copied passages roughly two times shorter than prior work, and learns to compress and merge sentences without supervision.

This paper describes the Critical Role Dungeons and Dragons Dataset (CRD3) and related analyses. Critical Role is an unscripted, live-streamed show where a fixed group of people play Dungeons and Dragons, an open-ended role-playing game. The dataset is collected from 159 Critical Role episodes transcribed to text dialogues, consisting of 398,682 turns. It also includes corresponding abstractive summaries collected from the Fandom wiki. The dataset is linguistically unique in that the narratives are generated entirely through player collaboration and spoken interaction. For each dialogue, there are a large number of turns, multiple abstractive summaries with varying levels of detail, and semantic ties to the previous dialogues. In addition, we provide a data augmentation method that produces 34,243 summary-dialogue chunk pairs to support current neural ML approaches, and we provide an abstractive summarization benchmark and evaluation.