How Do We Answer Complex Questions: Discourse Structure of Long-form Answers

Long-form answers, consisting of multiple sentences, can provide nuanced and comprehensive answers to a broader set of questions. To better understand this complex and understudied task, we study the functional structure of long-form answers collected from three datasets, ELI5, WebGPT and Natural Questions. Our main goal is to understand how humans organize information to craft complex answers. We develop an ontology of six sentence-level functional roles for long-form answers, and annotate 3.9k sentences in 640 answer paragraphs. Different answer collection methods manifest in different discourse structures. We further analyze model-generated answers -- finding that annotators agree less with each other when annotating model-generated answers compared to annotating human-written answers. Our annotated data enables training a strong classifier that can be used for automatic analysis. We hope our work can inspire future research on discourse-level modeling and evaluation of long-form QA systems.

Discourse Comprehension: A Question Answering Framework to Represent Sentence Connections

While there has been substantial progress in text comprehension through simple factoid question answering, more holistic comprehension of a discourse still presents a major challenge. Someone critically reflecting on a text as they read it will pose curiosity-driven, often open-ended questions, which reflect deep understanding of the content and require complex reasoning to answer. A key challenge in building and evaluating models for this type of discourse comprehension is the lack of annotated data, especially since finding answers to such questions (which may not be answered at all) requires high cognitive load for annotators over long documents. This paper presents a novel paradigm that enables scalable data collection targeting the comprehension of news documents, viewing these questions through the lens of discourse. The resulting corpus, DCQA (Discourse Comprehension by Question Answering), consists of 22,430 question-answer pairs across 607 English documents. DCQA captures both discourse and semantic links between sentences in the form of free-form, open-ended questions. On an evaluation set that we annotated on questions from the INQUISITIVE dataset, we show that DCQA provides valuable supervision for answering open-ended questions. We additionally design pre-training methods utilizing existing question-answering resources, and use synthetic data to accommodate unanswerable questions.

Training Dynamics for Text Summarization Models

Pre-trained language models (e.g. BART) have shown impressive results when fine-tuned on large summarization datasets. However, little is understood about this fine-tuning process, including what knowledge is retained from pre-training models or how content selection and generation strategies are learnt across iterations. In this work, we analyze the training dynamics for generation models, focusing on news summarization. Across different datasets (CNN/DM, XSum, MediaSum) and summary properties, such as abstractiveness and hallucination, we study what the model learns at different stages of its fine-tuning process. We find that properties such as copy behavior are learnt earlier in the training process and these observations are robust across domains. On the other hand, factual errors, such as hallucination of unsupported facts, are learnt in the later stages, and this behavior is more varied across domains. Based on these observations, we explore complementary approaches for modifying training: first, disregarding high-loss tokens that are challenging to learn and second, disregarding low-loss tokens that are learnt very quickly. This simple training modification allows us to configure our model to achieve different goals, such as improving factuality or improving abstractiveness.

AES Systems Are Both Overstable And Oversensitive: Explaining Why And Proposing Defenses

Deep-learning based Automatic Essay Scoring (AES) systems are being actively used by states and language testing agencies alike to evaluate millions of candidates for life-changing decisions ranging from college applications to visa approvals. However, little research has been put to understand and interpret the black-box nature of deep-learning based scoring algorithms. Previous studies indicate that scoring models can be easily fooled. In this paper, we explore the reason behind their surprising adversarial brittleness. We utilize recent advances in interpretability to find the extent to which features such as coherence, content, vocabulary, and relevance are important for automated scoring mechanisms. We use this to investigate the oversensitivity i.e., large change in output score with a little change in input essay content) and overstability i.e., little change in output scores with large changes in input essay content) of AES. Our results indicate that autoscoring models, despite getting trained as "end-to-end" models with rich contextual embeddings such as BERT, behave like bag-of-words models. A few words determine the essay score without the requirement of any context making the model largely overstable. This is in stark contrast to recent probing studies on pre-trained representation learning models, which show that rich linguistic features such as parts-of-speech and morphology are encoded by them. Further, we also find that the models have learnt dataset biases, making them oversensitive. To deal with these issues, we propose detection-based protection models that can detect oversensitivity and overstability causing samples with high accuracies. We find that our proposed models are able to detect unusual attribution patterns and flag adversarial samples successfully.

Learning to Describe Solutions for Bug Reports Based on Developer Discussions

When a software bug is reported, developers engage in a discussion to collaboratively resolve it. While the solution is likely formulated within the discussion, it is often buried in a large amount of text, making it difficult to comprehend, which delays its implementation. To expedite bug resolution, we propose generating a concise natural language description of the solution by synthesizing relevant content within the discussion, which encompasses both natural language and source code. Furthermore, to support generating an informative description during an ongoing discussion, we propose a secondary task of determining when sufficient context about the solution emerges in real-time. We construct a dataset for these tasks with a novel technique for obtaining noisy supervision from repository changes linked to bug reports. We establish baselines for generating solution descriptions, and develop a classifier which makes a prediction following each new utterance on whether or not the necessary context for performing generation is available. Through automated and human evaluation, we find these tasks to form an ideal testbed for complex reasoning in long, bimodal dialogue context.

AES Are Both Overstable And Oversensitive: Explaining Why And Proposing Defenses

Deep-learning based Automatic Essay Scoring (AES) systems are being actively used by states and language testing agencies alike to evaluate millions of candidates for life-changing decisions ranging from college applications to visa approvals. However, little research has been put to understand and interpret the black-box nature of deep-learning based scoring algorithms. Previous studies indicate that scoring models can be easily fooled. In this paper, we explore the reason behind their surprising adversarial brittleness. We utilize recent advances in interpretability to find the extent to which features such as coherence, content, vocabulary, and relevance are important for automated scoring mechanisms. We use this to investigate the oversensitivity i.e., large change in output score with a little change in input essay content) and overstability i.e., little change in output scores with large changes in input essay content) of AES. Our results indicate that autoscoring models, despite getting trained as "end-to-end" models with rich contextual embeddings such as BERT, behave like bag-of-words models. A few words determine the essay score without the requirement of any context making the model largely overstable. This is in stark contrast to recent probing studies on pre-trained representation learning models, which show that rich linguistic features such as parts-of-speech and morphology are encoded by them. Further, we also find that the models have learnt dataset biases, making them oversensitive. To deal with these issues, we propose detection-based protection models that can detect oversensitivity and overstability causing samples with high accuracies. We find that our proposed models are able to detect unusual attribution patterns and flag adversarial samples successfully.

The Case for Claim Difficulty Assessment in Automatic Fact Checking

Fact-checking is the process (human, automated, or hybrid) by which claims (i.e., purported facts) are evaluated for veracity. In this article, we raise an issue that has received little attention in prior work - that some claims are far more difficult to fact-check than others. We discuss the implications this has for both practical fact-checking and research on automated fact-checking, including task formulation and dataset design. We report a manual analysis undertaken to explore factors underlying varying claim difficulty and categorize several distinct types of difficulty. We argue that prediction of claim difficulty is a missing component of today's automated fact-checking architectures, and we describe how this difficulty prediction task might be split into a set of distinct subtasks.

Evaluation Methodologies for Code Learning Tasks

There has been a growing interest in developing machine learning (ML) models for code learning tasks, e.g., comment generation and method naming. Despite substantial increase in the effectiveness of ML models, the evaluation methodologies, i.e., the way people split datasets into training, validation, and testing sets, were not well designed. Specifically, no prior work on the aforementioned topics considered the timestamps of code and comments during evaluation (e.g., examples in the testing set might be from 2010 and examples from the training set might be from 2020). This may lead to evaluations that are inconsistent with the intended use cases of the ML models. In this paper, we formalize a novel time-segmented evaluation methodology, as well as the two methodologies commonly used in the literature: mixed-project and cross-project. We argue that time-segmented methodology is the most realistic. We also describe various use cases of ML models and provide a guideline for using methodologies to evaluate each use case. To assess the impact of methodologies, we collect a dataset of code-comment pairs with timestamps to train and evaluate several recent code learning ML models for the comment generation and method naming tasks. Our results show that different methodologies can lead to conflicting and inconsistent results. We invite the community to adopt the time-segmented evaluation methodology.

When a crisis strikes: Emotion analysis and detection during COVID-19

Crises such as natural disasters, global pandemics, and social unrest continuously threaten our world and emotionally affect millions of people worldwide in distinct ways. Understanding emotions that people express during large-scale crises helps inform policy makers and first responders about the emotional states of the population as well as provide emotional support to those who need such support. We present CovidEmo, ~1K tweets labeled with emotions. We examine how well large pre-trained language models generalize across domains and crises in the task of perceived emotion prediction in the context of COVID-19. Our results show that existing models do not directly transfer from one disaster type to another but using labeled emotional corpora for domain adaptation is beneficial.