In-context learning (ICL), the ability of large language models to perform novel tasks by conditioning on a prompt with a few task examples, requires these examples to be informative about the test instance. The standard approach of independently ranking and selecting the most similar examples selects redundant examples while omitting important information. In this work, we show that BERTScore-Recall (BSR) selects better examples that demonstrate more of the salient aspects, e.g. reasoning patterns, of the test input. We further extend BSR and many standard metrics to easily optimizable set-level metrics, giving still better coverage of those salient aspects. On 15 datasets spanning 6 tasks and with 7 diverse LLMs, we show that (1) BSR is the superior metric for in-context example selection across the board, and (2) for compositional tasks, set selection using Set-BSR outperforms independent ranking by up to 17 points on average and, despite being training-free, surpasses methods that leverage task or LLM-specific training.

Content Warning: This paper contains examples of misgendering and erasure that could be offensive and potentially triggering. Gender bias in language technologies has been widely studied, but research has mostly been restricted to a binary paradigm of gender. It is essential also to consider non-binary gender identities, as excluding them can cause further harm to an already marginalized group. In this paper, we comprehensively evaluate popular language models for their ability to correctly use English gender-neutral pronouns (e.g., singular they, them) and neo-pronouns (e.g., ze, xe, thon) that are used by individuals whose gender identity is not represented by binary pronouns. We introduce Misgendered, a framework for evaluating large language models’ ability to correctly use preferred pronouns, consisting of (i) instances declaring an individual’s pronoun, followed by a sentence with a missing pronoun, and (ii) an experimental setup for evaluating masked and auto-regressive language models using a unified method. When prompted out-of-the-box, language models perform poorly at correctly predicting neo-pronouns (averaging 7.6% accuracy) and gender-neutral pronouns (averaging 31.0% accuracy). This inability to generalize results from a lack of representation of non-binary pronouns in training data and memorized associations. Few-shot adaptation with explicit examples in the prompt improves the performance but plateaus at only 45.4% for neo-pronouns. We release the full dataset, code, and demo at https://tamannahossainkay.github.io/misgendered/.

Recent advances in open-domain question answering (ODQA) have demonstrated impressive accuracy on general-purpose domains like Wikipedia. While some work has been investigating how well ODQA models perform when tested for out-of-domain (OOD) generalization, these studies have been conducted only under conservative shifts in data distribution and typically focus on a single component (i.e., retriever or reader) rather than an end-to-end system. This work proposes a more realistic end-to-end domain shift evaluation setting covering five diverse domains. We not only find that end-to-end models fail to generalize but that high retrieval scores often still yield poor answer prediction accuracy. To address these failures, we investigate several interventions, in the form of data augmentations, for improving model adaption and use our evaluation set to elucidate the relationship between the efficacy of an intervention scheme and the particular type of dataset shifts we consider. We propose a generalizability test that estimates the type of shift in a target dataset without training a model in the target domain and that the type of shift is predictive of which data augmentation schemes will be effective for domain adaption. Overall, we find that these interventions increase end-to-end performance by up to ~24 points.

Training a referring expression comprehension (ReC) model for a new visual domain requires collecting referring expressions, and potentially corresponding bounding boxes, for images in the domain. While large-scale pre-trained models are useful for image classification across domains, it remains unclear if they can be applied in a zero-shot manner to more complex tasks like ReC. We present ReCLIP, a simple but strong zero-shot baseline that repurposes CLIP, a state-of-the-art large-scale model, for ReC. Motivated by the close connection between ReC and CLIP’s contrastive pre-training objective, the first component of ReCLIP is a region-scoring method that isolates object proposals via cropping and blurring, and passes them to CLIP. However, through controlled experiments on a synthetic dataset, we find that CLIP is largely incapable of performing spatial reasoning off-the-shelf. We reduce the gap between zero-shot baselines from prior work and supervised models by as much as 29% on RefCOCOg, and on RefGTA (video game imagery), ReCLIP’s relative improvement over supervised ReC models trained on real images is 8%.

The ability to efficiently learn from little-to-no data is critical to applying NLP to tasks where data collection is costly or otherwise difficult. This is a challenging setting both academically and practically—particularly because training neutral models typically require large amount of labeled data. More recently, advances in pretraining on unlabelled data have brought up the potential of better zero-shot or few-shot learning (Devlin et al., 2019; Brown et al., 2020). In particular, over the past year, a great deal of research has been conducted to better learn from limited data using large-scale language models. In this tutorial, we aim at bringing interested NLP researchers up to speed about the recent and ongoing techniques for zero- and few-shot learning with pretrained language models. Additionally, our goal is to reveal new research opportunities to the audience, which will hopefully bring us closer to address existing challenges in this domain.

Training the deep neural networks that dominate NLP requires large datasets. These are often collected automatically or via crowdsourcing, and may exhibit systematic biases or annotation artifacts. By the latter we mean spurious correlations between inputs and outputs that do not represent a generally held causal relationship between features and classes; models that exploit such correlations may appear to perform a given task well, but fail on out of sample data. In this paper, we evaluate use of different attribution methods for aiding identification of training data artifacts. We propose new hybrid approaches that combine saliency maps (which highlight important input features) with instance attribution methods (which retrieve training samples influential to a given prediction). We show that this proposed training-feature attribution can be used to efficiently uncover artifacts in training data when a challenging validation set is available. We also carry out a small user study to evaluate whether these methods are useful to NLP researchers in practice, with promising results. We make code for all methods and experiments in this paper available.

Prompting language models (LMs) with training examples and task descriptions has been seen as critical to recent successes in few-shot learning. In this work, we show that finetuning LMs in the few-shot setting can considerably reduce the need for prompt engineering. In fact, one can use null prompts, prompts that contain neither task-specific templates nor training examples, and achieve competitive accuracy to manually-tuned prompts across a wide range of tasks. While finetuning LMs does introduce new parameters for each downstream task, we show that this memory overhead can be substantially reduced: finetuning only the bias terms can achieve comparable or better accuracy than standard finetuning while only updating 0.1% of the parameters. All in all, we recommend finetuning LMs for few-shot learning as it is more accurate, robust to different prompts, and can be made nearly as efficient as using frozen LMs.

Pretrained Language Models (LMs) have demonstrated ability to perform numerical reasoning by extrapolating from a few examples in few-shot settings. However, the extent to which this extrapolation relies on robust reasoning is unclear. In this paper, we investigate how well these models reason with terms that are less frequent in the pretraining data. In particular, we examine the correlations between the model performance on test instances and the frequency of terms from those instances in the pretraining data. We measure the strength of this correlation for a number of GPT-based language models (pretrained on the Pile dataset) on various numerical deduction tasks (e.g., arithmetic and unit conversion). Our results consistently demonstrate that models are more accurate on instances whose terms are more prevalent, in some cases above 70% (absolute) more accurate on the top 10% frequent terms in comparison to the bottom 10%. Overall, although LMs appear successful at few-shot numerical reasoning, our results raise the question of how much models actually generalize beyond pretraining data, and we encourage researchers to take the pretraining data into account when interpreting evaluation results.

A growing body of research has demonstrated the inability of NLP models to generalize compositionally and has tried to alleviate it through specialized architectures, training schemes, and data augmentation, among other approaches. In this work, we study a different approach: training on instances with diverse structures. We propose a model-agnostic algorithm for subsampling such sets of instances from a labeled instance pool with structured outputs. Evaluating on both compositional template splits and traditional IID splits of 5 semantic parsing datasets of varying complexity, we show that structurally diverse training using our algorithm leads to comparable or better generalization than prior algorithms in 9 out of 10 dataset-split type pairs. In general, we find structural diversity to consistently improve sample efficiency compared to random train sets. Moreover, we show that structurally diverse sampling yields comprehensive test sets that are a lot more challenging than IID test sets. Finally, we provide two explanations for improved generalization from diverse train sets: 1) improved coverage of output substructures, and 2) a reduction in spurious correlations between these substructures.

Fine-tuning continuous prompts for target tasks has recently emerged as a compact alternative to full model fine-tuning. Motivated by these promising results, we investigate the feasibility of extracting a discrete (textual) interpretation of continuous prompts that is faithful to the problem they solve. In practice, we observe a “wayward” behavior between the task solved by continuous prompts and their nearest neighbor discrete projections: We can find continuous prompts that solve a task while being projected to an arbitrary text (e.g., definition of a different or even a contradictory task), while being within a very small (2%) margin of the best continuous prompt of the same size for the task. We provide intuitions behind this odd and surprising behavior, as well as extensive empirical analyses quantifying the effect of various parameters. For instance, for larger model sizes we observe higher waywardness, i.e, we can find prompts that more closely map to any arbitrary text with a smaller drop in accuracy. These findings have important implications relating to the difficulty of faithfully interpreting continuous prompts and their generalization across models and tasks, providing guidance for future progress in prompting language models.

Textual knowledge bases such as Wikipedia require considerable effort to keep up to date and consistent. While automated writing assistants could potentially ease this burden, the problem of suggesting edits grounded in external knowledge has been under-explored. In this paper, we introduce the novel generation task of *faithfully reflecting updated information in text* (FRUIT) where the goal is to update an existing article given new evidence. We release the FRUIT-WIKI dataset, a collection of over 170K distantly supervised data produced from pairs of Wikipedia snapshots, along with our data generation pipeline and a gold evaluation set of 914 instances whose edits are guaranteed to be supported by the evidence. We provide benchmark results for popular generation systems as well as EDIT5 – a T5-based approach tailored to editing we introduce that establishes the state of the art. Our analysis shows that developing models that can update articles faithfully requires new capabilities for neural generation models, and opens doors to many new applications.

Answering complex questions that require making latent decisions is a challenging task, especially when limited supervision is available. Recent works leverage the capabilities of large language models (LMs) to perform complex question answering in a few-shot setting by demonstrating how to output intermediate rationalizations while solving the complex question in a single pass. We introduce “Successive Prompting” where, we iteratively break down a complex task into a simple task, solve it, and then repeat the process until we get the final solution. Successive prompting decouples the supervision for decomposing complex questions from the supervision for answering simple questions, allowing us to (1) have multiple opportunities to query in-context examples at each reasoning step (2) learn question decomposition separately from question answering, including using synthetic data, and (3) use bespoke (fine-tuned) components for reasoning steps where a large LM does not perform well. The intermediate supervision is typically manually written, which can be expensive to collect. We introduce a way to generate synthetic dataset which can be used to bootstrap model’s ability to decompose and answer intermediate questions. Our best model (with successive prompting) achieves an improvement in F1 of ~5% when compared with a state-of-the-art model with synthetic augmentations and few-shot version of the DROP dataset.

Recently introduced language model prompting methods can achieve high accuracy in zero- and few-shot settings while requiring few to no learned task-specific parameters. Nevertheless, these methods still often trail behind full model finetuning. In this work, we investigate if a dedicated continued pretraining stage could improve “promptability”, i.e., zero-shot performance with natural language prompts or few-shot performance with prompt tuning. We reveal settings where existing continued pretraining methods lack promptability. We also identify current methodological gaps, which we fill with thorough large-scale experiments. We demonstrate that a simple recipe, continued pretraining that incorporates a trainable prompt during multi-task learning, leads to improved promptability in both zero- and few-shot settings compared to existing methods, up to 31% relative. On the other hand, we find that continued pretraining using MAML-style meta-learning, a method that directly optimizes few-shot promptability, yields subpar performance. We validate our findings with two prompt tuning methods, and, based on our results, we provide concrete recommendations to optimize promptability for different use cases.

Current evaluation schemes for large language models often fail to consider the impact of the overlap between pretraining corpus and test data on model performance statistics. Snoopy is an online interface that allows researchers to study this impact in few-shot learning settings. Our demo provides term frequency statistics for the Pile, which is an 800 GB corpus, accompanied by the precomputed performance of EleutherAI/GPT models on more than 20 NLP benchmarks, including numerical, commonsense reasoning, natural language understanding, and question-answering tasks. Snoopy allows a user to interactively align specific terms in test instances with their frequency in the Pile, enabling exploratory analysis of how term frequency is related to the accuracy of the models, which are hard to discover through automated means. A user can look at correlations over various model sizes and numbers of in-context examples and visualize the result across multiple (potentially aggregated) datasets. Using Snoopy, we show that a researcher can quickly replicate prior analyses for numerical tasks, while simultaneously allowing for much more expansive exploration that was previously challenging. Snoopy is available at https://nlp.ics.uci.edu/snoopy.

Adversarial attacks curated against NLP models are increasingly becoming practical threats. Although various methods have been developed to detect adversarial attacks, securing learning-based NLP systems in practice would require more than identifying and evading perturbed instances. To address these issues, we propose a new set of adversary identification tasks, Attacker Attribute Classification via Textual Analysis (AACTA), that attempts to obtain more detailed information about the attackers from adversarial texts. Specifically, given a piece of adversarial text, we hope to accomplish tasks such as localizing perturbed tokens, identifying the attacker’s access level to the target model, determining the evasion mechanism imposed, and specifying the perturbation type employed by the attacking algorithm. Our contributions are as follows: we formalize the task of classifying attacker attributes, and create a benchmark on various target models from sentiment classification and abuse detection domains. We show that signals from BERT models and target models can be used to train classifiers that reveal the properties of the attacking algorithms. We demonstrate that adversarial attacks leave interpretable traces in both feature spaces of pre-trained language models and target models, making AACTA a promising direction towards more trustworthy NLP systems.

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.

Widespread adoption of deep models has motivated a pressing need for approaches to interpret network outputs and to facilitate model debugging. Instance attribution methods constitute one means of accomplishing these goals by retrieving training instances that (may have) led to a particular prediction. Influence functions (IF; Koh and Liang 2017) provide machinery for doing this by quantifying the effect that perturbing individual train instances would have on a specific test prediction. However, even approximating the IF is computationally expensive, to the degree that may be prohibitive in many cases. Might simpler approaches (e.g., retrieving train examples most similar to a given test point) perform comparably? In this work, we evaluate the degree to which different potential instance attribution agree with respect to the importance of training samples. We find that simple retrieval methods yield training instances that differ from those identified via gradient-based methods (such as IFs), but that nonetheless exhibit desirable characteristics similar to more complex attribution methods. Code for all methods and experiments in this paper is available at: https://github.com/successar/instance_attributions_NLP.

Much recent work in NLP has documented dataset artifacts, bias, and spurious correlations between input features and output labels. However, how to tell which features have “spurious” instead of legitimate correlations is typically left unspecified. In this work we argue that for complex language understanding tasks, all simple feature correlations are spurious, and we formalize this notion into a class of problems which we call competency problems. For example, the word “amazing” on its own should not give information about a sentiment label independent of the context in which it appears, which could include negation, metaphor, sarcasm, etc. We theoretically analyze the difficulty of creating data for competency problems when human bias is taken into account, showing that realistic datasets will increasingly deviate from competency problems as dataset size increases. This analysis gives us a simple statistical test for dataset artifacts, which we use to show more subtle biases than were described in prior work, including demonstrating that models are inappropriately affected by these less extreme biases. Our theoretical treatment of this problem also allows us to analyze proposed solutions, such as making local edits to dataset instances, and to give recommendations for future data collection and model design efforts that target competency problems.

Compositional, structured models are appealing because they explicitly decompose problems and provide interpretable intermediate outputs that give confidence that the model is not simply latching onto data artifacts. Learning these models is challenging, however, because end-task supervision only provides a weak indirect signal on what values the latent decisions should take. This often results in the model failing to learn to perform the intermediate tasks correctly. In this work, we introduce a way to leverage paired examples that provide stronger cues for learning latent decisions. When two related training examples share internal substructure, we add an additional training objective to encourage consistency between their latent decisions. Such an objective does not require external supervision for the values of the latent output, or even the end task, yet provides an additional training signal to that provided by individual training examples themselves. We apply our method to improve compositional question answering using neural module networks on the DROP dataset. We explore three ways to acquire paired questions in DROP: (a) discovering naturally occurring paired examples within the dataset, (b) constructing paired examples using templates, and (c) generating paired examples using a question generation model. We empirically demonstrate that our proposed approach improves both in- and out-of-distribution generalization and leads to correct latent decision predictions.

Compositional reasoning tasks such as multi-hop question answering require models to learn how to make latent decisions using only weak supervision from the final answer. Crowdsourced datasets gathered for these tasks, however, often contain only a slice of the underlying task distribution, which can induce unanticipated biases such as shallow word overlap between the question and context. Recent works have shown that discriminative training results in models that exploit these underlying biases to achieve a better held-out performance, without learning the right way to reason. We propose a generative context selection model for multi-hop QA that reasons about how the given question could have been generated given a context pair and not just independent contexts. We show that on HotpotQA, while being comparable to the state-of-the-art answering performance, our proposed generative passage selection model has a better performance (4.9% higher than baseline) on adversarial held-out set which tests robustness of model’s multi-hop reasoning capabilities.

Knowledge-dependent tasks typically use two sources of knowledge: parametric, learned at training time, and contextual, given as a passage at inference time. To understand how models use these sources together, we formalize the problem of knowledge conflicts, where the contextual information contradicts the learned information. Analyzing the behaviour of popular models, we measure their over-reliance on memorized information (the cause of hallucinations), and uncover important factors that exacerbate this behaviour. Lastly, we propose a simple method to mitigate over-reliance on parametric knowledge, which minimizes hallucination, and improves out-of-distribution generalization by 4% - 7%. Our findings demonstrate the importance for practitioners to evaluate model tendency to hallucinate rather than read, and show that our mitigation strategy encourages generalization to evolving information (i.e. time-dependent queries). To encourage these practices, we have released our framework for generating knowledge conflicts.

When training most modern reading comprehension models, all the questions associated with a context are treated as being independent from each other. However, closely related questions and their corresponding answers are not independent, and leveraging these relationships could provide a strong supervision signal to a model. Drawing on ideas from contrastive estimation, we introduce several new supervision losses that compare question-answer scores across multiple related instances. Specifically, we normalize these scores across various neighborhoods of closely contrasting questions and/or answers, adding a cross entropy loss term in addition to traditional maximum likelihood estimation. Our techniques require bundles of related question-answer pairs, which we either mine from within existing data or create using automated heuristics. We empirically demonstrate the effectiveness of training with instance bundles on two datasets—HotpotQA and ROPES—showing up to 9% absolute gains in accuracy.

Recent studies show that many NLP systems are sensitive and vulnerable to a small perturbation of inputs and do not generalize well across different datasets. This lack of robustness derails the use of NLP systems in real-world applications. This tutorial aims at bringing awareness of practical concerns about NLP robustness. It targets NLP researchers and practitioners who are interested in building reliable NLP systems. In particular, we will review recent studies on analyzing the weakness of NLP systems when facing adversarial inputs and data with a distribution shift. We will provide the audience with a holistic view of 1) how to use adversarial examples to examine the weakness of NLP models and facilitate debugging; 2) how to enhance the robustness of existing NLP models and defense against adversarial inputs; and 3) how the consideration of robustness affects the real-world NLP applications used in our daily lives. We will conclude the tutorial by outlining future research directions in this area.

Retrieval is a core component for open-domain NLP tasks. In open-domain tasks, multiple entities can share a name, making disambiguation an inherent yet under-explored problem. We propose an evaluation benchmark for assessing the entity disambiguation capabilities of these retrievers, which we call Ambiguous Entity Retrieval (AmbER) sets. We define an AmbER set as a collection of entities that share a name along with queries about those entities. By covering the set of entities for polysemous names, AmbER sets act as a challenging test of entity disambiguation. We create AmbER sets for three popular open-domain tasks: fact checking, slot filling, and question answering, and evaluate a diverse set of retrievers. We find that the retrievers exhibit popularity bias, significantly under-performing on rarer entities that share a name, e.g., they are twice as likely to retrieve erroneous documents on queries for the less popular entity under the same name. These experiments on AmbER sets show their utility as an evaluation tool and highlight the weaknesses of popular retrieval systems.

Streaming cross document entity coreference (CDC) systems disambiguate mentions of named entities in a scalable manner via incremental clustering. Unlike other approaches for named entity disambiguation (e.g., entity linking), streaming CDC allows for the disambiguation of entities that are unknown at inference time. Thus, it is well-suited for processing streams of data where new entities are frequently introduced. Despite these benefits, this task is currently difficult to study, as existing approaches are either evaluated on datasets that are no longer available, or omit other crucial details needed to ensure fair comparison. In this work, we address this issue by compiling a large benchmark adapted from existing free datasets, and performing a comprehensive evaluation of a number of novel and existing baseline models. We investigate: how to best encode mentions, which clustering algorithms are most effective for grouping mentions, how models transfer to different domains, and how bounding the number of mentions tracked during inference impacts performance. Our results show that the relative performance of neural and feature-based mention encoders varies across different domains, and in most cases the best performance is achieved using a combination of both approaches. We also find that performance is minimally impacted by limiting the number of tracked mentions.

The predominant challenge in weakly supervised semantic parsing is that of spurious programs that evaluate to correct answers for the wrong reasons. Prior work uses elaborate search strategies to mitigate the prevalence of spurious programs; however, they typically consider only one input at a time. In this work we explore the use of consistency between the output programs for related inputs to reduce the impact of spurious programs. We bias the program search (and thus the model’s training signal) towards programs that map the same phrase in related inputs to the same sub-parts in their respective programs. Additionally, we study the importance of designing logical formalisms that facilitate this kind of consistency-based training. We find that a more consistent formalism leads to improved model performance even without consistency-based training. When combined together, these two insights lead to a 10% absolute improvement over the best prior result on the Natural Language Visual Reasoning dataset.

Building general reading comprehension systems, capable of solving multiple datasets at the same time, is a recent aspirational goal in the research community. Prior work has focused on model architecture or generalization to held out datasets, and largely passed over the particulars of the multi-task learning set up. We show that a simple dynamic sampling strategy, selecting instances for training proportional to the multi-task model’s current performance on a dataset relative to its single task performance, gives substantive gains over prior multi-task sampling strategies, mitigating the catastrophic forgetting that is common in multi-task learning. We also demonstrate that allowing instances of different tasks to be interleaved as much as possible between each epoch and batch has a clear benefit in multitask performance over forcing task homogeneity at the epoch or batch level. Our final model shows greatly increased performance over the best model on ORB, a recently-released multitask reading comprehension benchmark.

Language models that use additional latent structures (e.g., syntax trees, coreference chains, knowledge graph links) provide several advantages over traditional language models. However, likelihood-based evaluation of these models is often intractable as it requires marginalizing over the latent space. Existing works avoid this issue by using importance sampling. Although this approach has asymptotic guarantees, analysis is rarely conducted on the effect of decisions such as sample size and choice of proposal distribution on the reported estimates. In this paper, we carry out this analysis for three models: RNNG, EntityNLM, and KGLM. In addition, we elucidate subtle differences in how importance sampling is applied in these works that can have substantial effects on the final estimates, as well as provide theoretical results which reinforce the validity of this technique.

Although measuring held-out accuracy has been the primary approach to evaluate generalization, it often overestimates the performance of NLP models, while alternative approaches for evaluating models either focus on individual tasks or on specific behaviors. Inspired by principles of behavioral testing in software engineering, we introduce CheckList, a task-agnostic methodology for testing NLP models. CheckList includes a matrix of general linguistic capabilities and test types that facilitate comprehensive test ideation, as well as a software tool to generate a large and diverse number of test cases quickly. We illustrate the utility of CheckList with tests for three tasks, identifying critical failures in both commercial and state-of-art models. In a user study, a team responsible for a commercial sentiment analysis model found new and actionable bugs in an extensively tested model. In another user study, NLP practitioners with CheckList created twice as many tests, and found almost three times as many bugs as users without it.

Neural module networks (NMNs) are a popular approach for modeling compositionality: they achieve high accuracy when applied to problems in language and vision, while reflecting the compositional structure of the problem in the network architecture. However, prior work implicitly assumed that the structure of the network modules, describing the abstract reasoning process, provides a faithful explanation of the model’s reasoning; that is, that all modules perform their intended behaviour. In this work, we propose and conduct a systematic evaluation of the intermediate outputs of NMNs on NLVR2 and DROP, two datasets which require composing multiple reasoning steps. We find that the intermediate outputs differ from the expected output, illustrating that the network structure does not provide a faithful explanation of model behaviour. To remedy that, we train the model with auxiliary supervision and propose particular choices for module architecture that yield much better faithfulness, at a minimal cost to accuracy.

Complex compositional reading comprehension datasets require performing latent sequential decisions that are learned via supervision from the final answer. A large combinatorial space of possible decision paths that result in the same answer, compounded by the lack of intermediate supervision to help choose the right path, makes the learning particularly hard for this task. In this work, we study the benefits of collecting intermediate reasoning supervision along with the answer during data collection. We find that these intermediate annotations can provide two-fold benefits. First, we observe that for any collection budget, spending a fraction of it on intermediate annotations results in improved model performance, for two complex compositional datasets: DROP and Quoref. Second, these annotations encourage the model to learn the correct latent reasoning steps, helping combat some of the biases introduced during the data collection process.

The question of the utility of the blind peer-review system is fundamental to scientific research. Some studies investigate exactly how “blind” the papers are in the double-blind review system by manually or automatically identifying the true authors, mainly suggesting the number of self-citations in the submitted manuscripts as the primary signal for identity. However, related work on the automated approaches are limited by the sizes of their datasets and the restricted experimental setup, thus they lack practical insights into the blind review process. In this work, we train models that identify the authors, their affiliations, and their nationalities through real-world, large-scale experiments on the Microsoft Academic Graph, including the cold start scenario. Our models are accurate; we identify at least one of authors, affiliations, and nationalities of held-out papers with 40.3%, 47.9% and 86.0% accuracy respectively, from the top-10 guesses of our models. However, through insights from the model, we demonstrate that these entities are identifiable with a small number of guesses primarily by using a combination of self-citations, social, and common citations. Moreover, our further analysis on the results leads to interesting findings, such as that prominent affiliations are easily identifiable (e.g. 93.8% of test papers written by Microsoft are identified with top-10 guesses). The experimental results show, against conventional belief, that the self-citations are no more informative than looking at the common citations, thus suggesting that removing self-citations is not sufficient for authors to maintain their anonymity.

To identify what entities are being talked about in tweets, we need to automatically link named entities that appear in tweets to structured KBs like WikiData. Existing approaches often struggle with such short, noisy texts, or their complex design and reliance on supervision make them brittle, difficult to use and maintain, and lose significance over time. Further, there is a lack of a large, linked corpus of tweets to aid researchers, along with lack of gold dataset to evaluate the accuracy of entity linking. In this paper, we introduce (1) Tweeki, an unsupervised, modular entity linking system for Twitter, (2) TweekiData, a large, automatically-annotated corpus of Tweets linked to entities in WikiData, and (3) TweekiGold, a gold dataset for entity linking evaluation. Through comprehensive analysis, we show that Tweeki is comparable to the performance of recent state-of-the-art entity linkers models, the dataset is of high quality, and a use case of how the dataset can be used to improve downstream tasks in social media analysis (geolocation prediction).

Gradient-based analysis methods, such as saliency map visualizations and adversarial input perturbations, have found widespread use in interpreting neural NLP models due to their simplicity, flexibility, and most importantly, the fact that they directly reflect the model internals. In this paper, however, we demonstrate that the gradients of a model are easily manipulable, and thus bring into question the reliability of gradient-based analyses. In particular, we merge the layers of a target model with a Facade Model that overwhelms the gradients without affecting the predictions. This Facade Model can be trained to have gradients that are misleading and irrelevant to the task, such as focusing only on the stop words in the input. On a variety of NLP tasks (sentiment analysis, NLI, and QA), we show that the merged model effectively fools different analysis tools: saliency maps differ significantly from the original model’s, input reduction keeps more irrelevant input tokens, and adversarial perturbations identify unimportant tokens as being highly important.

Standard test sets for supervised learning evaluate in-distribution generalization. Unfortunately, when a dataset has systematic gaps (e.g., annotation artifacts), these evaluations are misleading: a model can learn simple decision rules that perform well on the test set but do not capture the abilities a dataset is intended to test. We propose a more rigorous annotation paradigm for NLP that helps to close systematic gaps in the test data. In particular, after a dataset is constructed, we recommend that the dataset authors manually perturb the test instances in small but meaningful ways that (typically) change the gold label, creating contrast sets. Contrast sets provide a local view of a model’s decision boundary, which can be used to more accurately evaluate a model’s true linguistic capabilities. We demonstrate the efficacy of contrast sets by creating them for 10 diverse NLP datasets (e.g., DROP reading comprehension, UD parsing, and IMDb sentiment analysis). Although our contrast sets are not explicitly adversarial, model performance is significantly lower on them than on the original test sets—up to 25% in some cases. We release our contrast sets as new evaluation benchmarks and encourage future dataset construction efforts to follow similar annotation processes.

Understanding the relationship between figures and text is key to scientific document understanding. Medical figures in particular are quite complex, often consisting of several subfigures (75% of figures in our dataset), with detailed text describing their content. Previous work studying figures in scientific papers focused on classifying figure content rather than understanding how images relate to the text. To address challenges in figure retrieval and figure-to-text alignment, we introduce MedICaT, a dataset of medical images in context. MedICaT consists of 217K images from 131K open access biomedical papers, and includes captions, inline references for 74% of figures, and manually annotated subfigures and subcaptions for a subset of figures. Using MedICaT, we introduce the task of subfigure to subcaption alignment in compound figures and demonstrate the utility of inline references in image-text matching. Our data and code can be accessed at https://github.com/allenai/medicat.

The ongoing pandemic has heightened the need for developing tools to flag COVID-19-related misinformation on the internet, specifically on social media such as Twitter. However, due to novel language and the rapid change of information, existing misinformation detection datasets are not effective for evaluating systems designed to detect misinformation on this topic. Misinformation detection can be divided into two sub-tasks: (i) retrieval of misconceptions relevant to posts being checked for veracity, and (ii) stance detection to identify whether the posts Agree, Disagree, or express No Stance towards the retrieved misconceptions. To facilitate research on this task, we release COVIDLies (https://ucinlp.github.io/covid19 ), a dataset of 6761 expert-annotated tweets to evaluate the performance of misinformation detection systems on 86 different pieces of COVID-19 related misinformation. We evaluate existing NLP systems on this dataset, providing initial benchmarks and identifying key challenges for future models to improve upon.

The remarkable success of pretrained language models has motivated the study of what kinds of knowledge these models learn during pretraining. Reformulating tasks as fill-in-the-blanks problems (e.g., cloze tests) is a natural approach for gauging such knowledge, however, its usage is limited by the manual effort and guesswork required to write suitable prompts. To address this, we develop AutoPrompt, an automated method to create prompts for a diverse set of tasks, based on a gradient-guided search. Using AutoPrompt, we show that masked language models (MLMs) have an inherent capability to perform sentiment analysis and natural language inference without additional parameters or finetuning, sometimes achieving performance on par with recent state-of-the-art supervised models. We also show that our prompts elicit more accurate factual knowledge from MLMs than the manually created prompts on the LAMA benchmark, and that MLMs can be used as relation extractors more effectively than supervised relation extraction models. These results demonstrate that automatically generated prompts are a viable parameter-free alternative to existing probing methods, and as pretrained LMs become more sophisticated and capable, potentially a replacement for finetuning.

Posing reading comprehension as a generation problem provides a great deal of flexibility, allowing for open-ended questions with few restrictions on possible answers. However, progress is impeded by existing generation metrics, which rely on token overlap and are agnostic to the nuances of reading comprehension. To address this, we introduce a benchmark for training and evaluating generative reading comprehension metrics: MOdeling Correctness with Human Annotations. MOCHA contains 40K human judgement scores on model outputs from 6 diverse question answering datasets and an additional set of minimal pairs for evaluation. Using MOCHA, we train a Learned Evaluation metric for Reading Comprehension, LERC, to mimic human judgement scores. LERC outperforms baseline metrics by 10 to 36 absolute Pearson points on held-out annotations. When we evaluate robustness on minimal pairs, LERC achieves 80% accuracy, outperforming baselines by 14 to 26 absolute percentage points while leaving significant room for improvement. MOCHA presents a challenging problem for developing accurate and robust generative reading comprehension metrics.

Although neural NLP models are highly expressive and empirically successful, they also systematically fail in counterintuitive ways and are opaque in their decision-making process. This tutorial will provide a background on interpretation techniques, i.e., methods for explaining the predictions of NLP models. We will first situate example-specific interpretations in the context of other ways to understand models (e.g., probing, dataset analyses). Next, we will present a thorough study of example-specific interpretations, including saliency maps, input perturbations (e.g., LIME, input reduction), adversarial attacks, and influence functions. Alongside these descriptions, we will walk through source code that creates and visualizes interpretations for a diverse set of NLP tasks. Finally, we will discuss open problems in the field, e.g., evaluating, extending, and improving interpretation methods.

Multi-hop reading comprehension (RC) questions are challenging because they require reading and reasoning over multiple paragraphs. We argue that it can be difficult to construct large multi-hop RC datasets. For example, even highly compositional questions can be answered with a single hop if they target specific entity types, or the facts needed to answer them are redundant. Our analysis is centered on HotpotQA, where we show that single-hop reasoning can solve much more of the dataset than previously thought. We introduce a single-hop BERT-based RC model that achieves 67 F1—comparable to state-of-the-art multi-hop models. We also design an evaluation setting where humans are not shown all of the necessary paragraphs for the intended multi-hop reasoning but can still answer over 80% of questions. Together with detailed error analysis, these results suggest there should be an increasing focus on the role of evidence in multi-hop reasoning and possibly even a shift towards information retrieval style evaluations with large and diverse evidence collections.

Modeling human language requires the ability to not only generate fluent text but also encode factual knowledge. However, traditional language models are only capable of remembering facts seen at training time, and often have difficulty recalling them. To address this, we introduce the knowledge graph language model (KGLM), a neural language model with mechanisms for selecting and copying facts from a knowledge graph that are relevant to the context. These mechanisms enable the model to render information it has never seen before, as well as generate out-of-vocabulary tokens. We also introduce the Linked WikiText-2 dataset, a corpus of annotated text aligned to the Wikidata knowledge graph whose contents (roughly) match the popular WikiText-2 benchmark. In experiments, we demonstrate that the KGLM achieves significantly better performance than a strong baseline language model. We additionally compare different language model’s ability to complete sentences requiring factual knowledge, showing that the KGLM outperforms even very large language models in generating facts.

Although current evaluation of question-answering systems treats predictions in isolation, we need to consider the relationship between predictions to measure true understanding. A model should be penalized for answering “no” to “Is the rose red?” if it answers “red” to “What color is the rose?”. We propose a method to automatically extract such implications for instances from two QA datasets, VQA and SQuAD, which we then use to evaluate the consistency of models. Human evaluation shows these generated implications are well formed and valid. Consistency evaluation provides crucial insights into gaps in existing models, while retraining with implication-augmented data improves consistency on both synthetic and human-generated implications.

Contextual word representations, typically trained on unstructured, unlabeled text, do not contain any explicit grounding to real world entities and are often unable to remember facts about those entities. We propose a general method to embed multiple knowledge bases (KBs) into large scale models, and thereby enhance their representations with structured, human-curated knowledge. For each KB, we first use an integrated entity linker to retrieve relevant entity embeddings, then update contextual word representations via a form of word-to-entity attention. In contrast to previous approaches, the entity linkers and self-supervised language modeling objective are jointly trained end-to-end in a multitask setting that combines a small amount of entity linking supervision with a large amount of raw text. After integrating WordNet and a subset of Wikipedia into BERT, the knowledge enhanced BERT (KnowBert) demonstrates improved perplexity, ability to recall facts as measured in a probing task and downstream performance on relationship extraction, entity typing, and word sense disambiguation. KnowBert’s runtime is comparable to BERT’s and it scales to large KBs.

Adversarial examples highlight model vulnerabilities and are useful for evaluation and interpretation. We define universal adversarial triggers: input-agnostic sequences of tokens that trigger a model to produce a specific prediction when concatenated to any input from a dataset. We propose a gradient-guided search over tokens which finds short trigger sequences (e.g., one word for classification and four words for language modeling) that successfully trigger the target prediction. For example, triggers cause SNLI entailment accuracy to drop from 89.94% to 0.55%, 72% of “why” questions in SQuAD to be answered “to kill american people”, and the GPT-2 language model to spew racist output even when conditioned on non-racial contexts. Furthermore, although the triggers are optimized using white-box access to a specific model, they transfer to other models for all tasks we consider. Finally, since triggers are input-agnostic, they provide an analysis of global model behavior. For instance, they confirm that SNLI models exploit dataset biases and help to diagnose heuristics learned by reading comprehension models.

The ability to understand and work with numbers (numeracy) is critical for many complex reasoning tasks. Currently, most NLP models treat numbers in text in the same way as other tokens—they embed them as distributed vectors. Is this enough to capture numeracy? We begin by investigating the numerical reasoning capabilities of a state-of-the-art question answering model on the DROP dataset. We find this model excels on questions that require numerical reasoning, i.e., it already captures numeracy. To understand how this capability emerges, we probe token embedding methods (e.g., BERT, GloVe) on synthetic list maximum, number decoding, and addition tasks. A surprising degree of numeracy is naturally present in standard embeddings. For example, GloVe and word2vec accurately encode magnitude for numbers up to 1,000. Furthermore, character-level embeddings are even more precise—ELMo captures numeracy the best for all pre-trained methods—but BERT, which uses sub-word units, is less exact.

Neural NLP models are increasingly accurate but are imperfect and opaque—they break in counterintuitive ways and leave end users puzzled at their behavior. Model interpretation methods ameliorate this opacity by providing explanations for specific model predictions. Unfortunately, existing interpretation codebases make it difficult to apply these methods to new models and tasks, which hinders adoption for practitioners and burdens interpretability researchers. We introduce AllenNLP Interpret, a flexible framework for interpreting NLP models. The toolkit provides interpretation primitives (e.g., input gradients) for any AllenNLP model and task, a suite of built-in interpretation methods, and a library of front-end visualization components. We demonstrate the toolkit’s flexibility and utility by implementing live demos for five interpretation methods (e.g., saliency maps and adversarial attacks) on a variety of models and tasks (e.g., masked language modeling using BERT and reading comprehension using BiDAF). These demos, alongside our code and tutorials, are available at https://allennlp.org/interpret.

As the complexity of question answering (QA) datasets evolve, moving away from restricted formats like span extraction and multiple-choice (MC) to free-form answer generation, it is imperative to understand how well current metrics perform in evaluating QA. This is especially important as existing metrics (BLEU, ROUGE, METEOR, and F1) are computed using n-gram similarity and have a number of well-known drawbacks. In this work, we study the suitability of existing metrics in QA. For generative QA, we show that while current metrics do well on existing datasets, converting multiple-choice datasets into free-response datasets is challenging for current metrics. We also look at span-based QA, where F1 is a reasonable metric. We show that F1 may not be suitable for all extractive QA tasks depending on the answer types. Our study suggests that while current metrics may be suitable for existing QA datasets, they limit the complexity of QA datasets that can be created. This is especially true in the context of free-form QA, where we would like our models to be able to generate more complex and abstractive answers, thus necessitating new metrics that go beyond n-gram based matching. As a step towards a better QA metric, we explore using BERTScore, a recently proposed metric for evaluating translation, for QA. We find that although it fails to provide stronger correlation with human judgements, future work focused on tailoring a BERT-based metric to QA evaluation may prove fruitful.

Reading comprehension is one of the crucial tasks for furthering research in natural language understanding. A lot of diverse reading comprehension datasets have recently been introduced to study various phenomena in natural language, ranging from simple paraphrase matching and entity typing to entity tracking and understanding the implications of the context. Given the availability of many such datasets, comprehensive and reliable evaluation is tedious and time-consuming for researchers working on this problem. We present an evaluation server, ORB, that reports performance on seven diverse reading comprehension datasets, encouraging and facilitating testing a single model’s capability in understanding a wide variety of reading phenomena. The evaluation server places no restrictions on how models are trained, so it is a suitable test bed for exploring training paradigms and representation learning for general reading facility. As more suitable datasets are released, they will be added to the evaluation server. We also collect and include synthetic augmentations for these datasets, testing how well models can handle out-of-domain questions.

We introduce entity post-modifier generation as an instance of a collaborative writing task. Given a sentence about a target entity, the task is to automatically generate a post-modifier phrase that provides contextually relevant information about the entity. For example, for the sentence, “Barack Obama, _______, supported the #MeToo movement.”, the phrase “a father of two girls” is a contextually relevant post-modifier. To this end, we build PoMo, a post-modifier dataset created automatically from news articles reflecting a journalistic need for incorporating entity information that is relevant to a particular news event. PoMo consists of more than 231K sentences with post-modifiers and associated facts extracted from Wikidata for around 57K unique entities. We use crowdsourcing to show that modeling contextual relevance is necessary for accurate post-modifier generation. We adapt a number of existing generation approaches as baselines for this dataset. Our results show there is large room for improvement in terms of both identifying relevant facts to include (knowing which claims are relevant gives a >20% improvement in BLEU score), and generating appropriate post-modifier text for the context (providing relevant claims is not sufficient for accurate generation). We conduct an error analysis that suggests promising directions for future research.

Reading comprehension has recently seen rapid progress, with systems matching humans on the most popular datasets for the task. However, a large body of work has highlighted the brittleness of these systems, showing that there is much work left to be done. We introduce a new reading comprehension benchmark, DROP, which requires Discrete Reasoning Over the content of Paragraphs. In this crowdsourced, adversarially-created, 55k-question benchmark, a system must resolve references in a question, perhaps to multiple input positions, and perform discrete operations over them (such as addition, counting, or sorting). These operations require a much more comprehensive understanding of the content of paragraphs, as they remove the paraphrase-and-entity-typing shortcuts available in prior datasets. We apply state-of-the-art methods from both the reading comprehension and semantic parsing literatures on this dataset and show that the best systems only achieve 38.4% F1 on our generalized accuracy metric, while expert human performance is 96%. We additionally present a new model that combines reading comprehension methods with simple numerical reasoning to achieve 51% F1.

Language is gendered if the context surrounding a mention is suggestive of a particular binary gender for that mention. Detecting the different ways in which language is gendered is an important task since gendered language can bias NLP models (such as for coreference resolution). This task is challenging since genderedness is often expressed in subtle ways. Existing approaches need considerable annotation efforts for each language, domain, and author, and often require handcrafted lexicons and features. Additionally, these approaches do not provide a quantifiable measure of how gendered the text is, nor are they applicable at the fine-grained mention level. In this paper, we use existing NLP pipelines to automatically annotate gender of mentions in the text. On corpora labeled using this method, we train a supervised classifier to predict the gender of any mention from its context and evaluate it on unseen text. The model confidence for a mention’s gender can be used as a proxy to indicate the level of genderedness of the context. We test this gendered language detector on movie summaries, movie reviews, news articles, and fiction novels, achieving an AUC-ROC of up to 0.71, and observe that the model predictions agree with human judgments collected for this task. We also provide examples of detected gendered sentences from aforementioned domains.

Representing entities and relations in an embedding space is a well-studied approach for machine learning on relational data. Existing approaches, however, primarily focus on improving accuracy and overlook other aspects such as robustness and interpretability. In this paper, we propose adversarial modifications for link prediction models: identifying the fact to add into or remove from the knowledge graph that changes the prediction for a target fact after the model is retrained. Using these single modifications of the graph, we identify the most influential fact for a predicted link and evaluate the sensitivity of the model to the addition of fake facts. We introduce an efficient approach to estimate the effect of such modifications by approximating the change in the embeddings when the knowledge graph changes. To avoid the combinatorial search over all possible facts, we train a network to decode embeddings to their corresponding graph components, allowing the use of gradient-based optimization to identify the adversarial modification. We use these techniques to evaluate the robustness of link prediction models (by measuring sensitivity to additional facts), study interpretability through the facts most responsible for predictions (by identifying the most influential neighbors), and detect incorrect facts in the knowledge base.

Adversarial learning is a game-theoretic learning paradigm, which has achieved huge successes in the field of Computer Vision recently. Adversarial learning is also a general framework that enables a variety of learning models, including the popular Generative Adversarial Networks (GANs). Due to the discrete nature of language, designing adversarial learning models is still challenging for NLP problems. In this tutorial, we provide a gentle introduction to the foundation of deep adversarial learning, as well as some practical problem formulations and solutions in NLP. We describe recent advances in deep adversarial learning for NLP, with a special focus on generation, adversarial examples & rules, and dialogue. We provide an overview of the research area, categorize different types of adversarial learning models, and discuss pros and cons, aiming at providing some practical perspectives on the future of adversarial learning for solving real-world NLP problems.

Most work in machine reading focuses on question answering problems where the answer is directly expressed in the text to read. However, many real-world question answering problems require the reading of text not because it contains the literal answer, but because it contains a recipe to derive an answer together with the reader’s background knowledge. One example is the task of interpreting regulations to answer “Can I...?” or “Do I have to...?” questions such as “I am working in Canada. Do I have to carry on paying UK National Insurance?” after reading a UK government website about this topic. This task requires both the interpretation of rules and the application of background knowledge. It is further complicated due to the fact that, in practice, most questions are underspecified, and a human assistant will regularly have to ask clarification questions such as “How long have you been working abroad?” when the answer cannot be directly derived from the question and text. In this paper, we formalise this task and develop a crowd-sourcing strategy to collect 37k task instances based on real-world rules and crowd-generated questions and scenarios. We analyse the challenges of this task and assess its difficulty by evaluating the performance of rule-based and machine-learning baselines. We observe promising results when no background knowledge is necessary, and substantial room for improvement whenever background knowledge is needed.

Representing entities and relations in an embedding space is a well-studied approach for machine learning on relational data. Existing approaches, however, primarily focus on simple link structure between a finite set of entities, ignoring the variety of data types that are often used in knowledge bases, such as text, images, and numerical values. In this paper, we propose multimodal knowledge base embeddings (MKBE) that use different neural encoders for this variety of observed data, and combine them with existing relational models to learn embeddings of the entities and multimodal data. Further, using these learned embedings and different neural decoders, we introduce a novel multimodal imputation model to generate missing multimodal values, like text and images, from information in the knowledge base. We enrich existing relational datasets to create two novel benchmarks that contain additional information such as textual descriptions and images of the original entities. We demonstrate that our models utilize this additional information effectively to provide more accurate link prediction, achieving state-of-the-art results with a considerable gap of 5-7% over existing methods. Further, we evaluate the quality of our generated multimodal values via a user study.

Complex machine learning models for NLP are often brittle, making different predictions for input instances that are extremely similar semantically. To automatically detect this behavior for individual instances, we present semantically equivalent adversaries (SEAs) – semantic-preserving perturbations that induce changes in the model’s predictions. We generalize these adversaries into semantically equivalent adversarial rules (SEARs) – simple, universal replacement rules that induce adversaries on many instances. We demonstrate the usefulness and flexibility of SEAs and SEARs by detecting bugs in black-box state-of-the-art models for three domains: machine comprehension, visual question-answering, and sentiment analysis. Via user studies, we demonstrate that we generate high-quality local adversaries for more instances than humans, and that SEARs induce four times as many mistakes as the bugs discovered by human experts. SEARs are also actionable: retraining models using data augmentation significantly reduces bugs, while maintaining accuracy.

For accurate entity linking, we need to capture various information aspects of an entity, such as its description in a KB, contexts in which it is mentioned, and structured knowledge. Additionally, a linking system should work on texts from different domains without requiring domain-specific training data or hand-engineered features. In this work we present a neural, modular entity linking system that learns a unified dense representation for each entity using multiple sources of information, such as its description, contexts around its mentions, and its fine-grained types. We show that the resulting entity linking system is effective at combining these sources, and performs competitively, sometimes out-performing current state-of-the-art systems across datasets, without requiring any domain-specific training data or hand-engineered features. We also show that our model can effectively “embed” entities that are new to the KB, and is able to link its mentions accurately.

We present a novel way for designing complex joint inference and learning models using Saul (Kordjamshidi et al., 2015), a recently-introduced declarative learning-based programming language (DeLBP). We enrich Saul with components that are necessary for a broad range of learning based Natural Language Processing tasks at various levels of granularity. We illustrate these advances using three different, well-known NLP problems, and show how these generic learning and inference modules can directly exploit Saul’s graph-based data representation. These properties allow the programmer to easily switch between different model formulations and configurations, and consider various kinds of dependencies and correlations among variables of interest with minimal programming effort. We argue that Saul provides an extremely useful paradigm both for the design of advanced NLP systems and for supporting advanced research in NLP.

Recent research on entity linking (EL) has introduced a plethora of promising techniques, ranging from deep neural networks to joint inference. But despite numerous papers there is surprisingly little understanding of the state of the art in EL. We attack this confusion by analyzing differences between several versions of the EL problem and presenting a simple yet effective, modular, unsupervised system, called Vinculum, for entity linking. We conduct an extensive evaluation on nine data sets, comparing Vinculum with two state-of-the-art systems, and elucidate key aspects of the system that include mention extraction, candidate generation, entity type prediction, entity coreference, and coherence.