Daniel S. Weld

Marissa Radensky, Simra Shahid, Raymond Fok et al. · allen-ai, uw

The scientific ideation process often involves blending facets of existing papers to create new ideas. We contribute Scideator, the first human-LLM system for facet-based scientific ideation. Starting from user-provided papers, Scideator extracts key facets -- purposes, mechanisms, and evaluations -- from these and related papers, allowing users to interactively recombine facets to synthesize ideas. Scideator is driven by three design choices: (1) human-in-the-loop facet recombination, in which users select facets from retrieved papers and the system generates ideas by finding analogies across them via the Faceted Idea Generator module; (2) distance-controlled retrieval via the Analogous Paper Facet Finder module, which surfaces papers ranging from the same topic to entirely different areas to provide a spectrum of directions; and (3) facet-based novelty verification via the Idea Novelty Checker module, a retrieve-then-rerank pipeline that helps users to evaluate idea originality using facets. In a user study with computer science researchers, Scideator provided significantly more creativity support than a baseline using the same backbone LLM without our facet-based modules, particularly in idea exploration and expressiveness. Ablations further show that the facets benefit the novelty checker: facet-based retrieve-then-rerank surfaces more relevant papers than standard retrieval and re-ranking, and a facet-grounded novelty classifier outperforms classifiers that reason over unstructured ideas and papers.

Rodney Kinney, Chloe Anastasiades, Russell Authur et al. · allen-ai, microsoft-research

The volume of scientific output is creating an urgent need for automated tools to help scientists keep up with developments in their field. Semantic Scholar (S2) is an open data platform and website aimed at accelerating science by helping scholars discover and understand scientific literature. We combine public and proprietary data sources using state-of-the-art techniques for scholarly PDF content extraction and automatic knowledge graph construction to build the Semantic Scholar Academic Graph, the largest open scientific literature graph to-date, with 200M+ papers, 80M+ authors, 550M+ paper-authorship edges, and 2.4B+ citation edges. The graph includes advanced semantic features such as structurally parsed text, natural language summaries, and vector embeddings. In this paper, we describe the components of the S2 data processing pipeline and the associated APIs offered by the platform. We will update this living document to reflect changes as we add new data offerings and improve existing services.

Tongshuang Wu, Hua Shen, Daniel S. Weld et al. · cmu, microsoft-research

The in-context learning capabilities of LLMs like GPT-3 allow annotators to customize an LLM to their specific tasks with a small number of examples. However, users tend to include only the most obvious patterns when crafting examples, resulting in underspecified in-context functions that fall short on unseen cases. Further, it is hard to know when "enough" examples have been included even for known patterns. In this work, we present ScatterShot, an interactive system for building high-quality demonstration sets for in-context learning. ScatterShot iteratively slices unlabeled data into task-specific patterns, samples informative inputs from underexplored or not-yet-saturated slices in an active learning manner, and helps users label more efficiently with the help of an LLM and the current example set. In simulation studies on two text perturbation scenarios, ScatterShot sampling improves the resulting few-shot functions by 4-5 percentage points over random sampling, with less variance as more examples are added. In a user study, ScatterShot greatly helps users in covering different patterns in the input space and labeling in-context examples more efficiently, resulting in better in-context learning and less user effort.

Daniel King, Zejiang Shen, Nishant Subramani et al. · allen-ai, cmu

Abstractive summarization systems today produce fluent and relevant output, but often "hallucinate" statements not supported by the source text. We analyze the connection between hallucinations and training data, and find evidence that models hallucinate because they train on target summaries that are unsupported by the source. Based on our findings, we present PINOCCHIO, a new decoding method that improves the consistency of a transformer-based abstractive summarizer by constraining beam search to avoid hallucinations. Given the model states and outputs at a given step, PINOCCHIO detects likely model hallucinations based on various measures of attribution to the source text. PINOCCHIO backtracks to find more consistent output, and can opt to produce no summary at all when no consistent generation can be found. In experiments, we find that PINOCCHIO improves the consistency of generation (in terms of F1) by an average of~67% on two abstractive summarization datasets.

Kyle Lo, Joseph Chee Chang, Andrew Head et al. · allen-ai, cmu

Scholarly publications are key to the transfer of knowledge from scholars to others. However, research papers are information-dense, and as the volume of the scientific literature grows, the need for new technology to support the reading process grows. In contrast to the process of finding papers, which has been transformed by Internet technology, the experience of reading research papers has changed little in decades. The PDF format for sharing research papers is widely used due to its portability, but it has significant downsides including: static content, poor accessibility for low-vision readers, and difficulty reading on mobile devices. This paper explores the question "Can recent advances in AI and HCI power intelligent, interactive, and accessible reading interfaces -- even for legacy PDFs?" We describe the Semantic Reader Project, a collaborative effort across multiple institutions to explore automatic creation of dynamic reading interfaces for research papers. Through this project, we've developed ten research prototype interfaces and conducted usability studies with more than 300 participants and real-world users showing improved reading experiences for scholars. We've also released a production reading interface for research papers that will incorporate the best features as they mature. We structure this paper around challenges scholars and the public face when reading research papers -- Discovery, Efficiency, Comprehension, Synthesis, and Accessibility -- and present an overview of our progress and remaining open challenges.

Tom Hope, Doug Downey, Oren Etzioni et al. · uw

We stand at the foot of a significant inflection in the trajectory of scientific discovery. As society continues on its fast-paced digital transformation, so does humankind's collective scientific knowledge and discourse. We now read and write papers in digitized form, and a great deal of the formal and informal processes of science are captured digitally -- including papers, preprints and books, code and datasets, conference presentations, and interactions in social networks and collaboration and communication platforms. The transition has led to the creation and growth of a tremendous amount of information -- much of which is available for public access -- opening exciting opportunities for computational models and systems that analyze and harness it. In parallel, exponential growth in data processing power has fueled remarkable advances in artificial intelligence, including large neural language models capable of learning powerful representations from unstructured text. Dramatic changes in scientific communication -- such as the advent of the first scientific journal in the 17th century -- have historically catalyzed revolutions in scientific thought. The confluence of societal and computational trends suggests that computer science is poised to ignite a revolution in the scientific process itself.

Marissa Radensky, Simra Shahid, Raymond Fok et al. · allen-ai, uw

The scientific ideation process often involves blending salient aspects of existing papers to create new ideas - a framework known as facet-based ideation. We contribute Scideator, the first human-LLM system for facet-based scientific ideation. Starting from a user-provided set of scientific papers, Scideator extracts key facets -- purposes, mechanisms, and evaluations -- from these and related papers, allowing users to explore the idea space by interactively recombining facets to synthesize inventive ideas. Scideator is driven by three design choices: (1) human-in-the-loop facet recombination, in which users select facets from retrieved papers and the system generates ideas by finding analogies across them via the Faceted Idea Generator module; (2) distance-controlled retrieval via the Analogous Paper Facet Finder module, which surfaces papers from the same topic to entirely different subareas to provide a spectrum of creative directions; and (3) facet-based novelty verification via the Idea Novelty Checker module, a retrieve-then-rerank pipeline that evaluates idea originality using facets. In a user study with computer science researchers, Scideator provided significantly more creativity support than a baseline using the same backbone LLM without our facet-based modules, particularly in idea exploration and expressiveness. Participants' favorite ideas more often included facets selected by themselves rather than the LLM, and participants used fewer free-text instructions with Scideator, indicating a preference for facet-level steering over prompting. Finally, re-ranking papers by facet matching rather than general relevance improved novelty classification accuracy from 13.79% to 89.66%.

Sonia K. Murthy, Kyle Lo, Daniel King et al. · allen-ai, uw

Systems that can automatically define unfamiliar terms hold the promise of improving the accessibility of scientific texts, especially for readers who may lack prerequisite background knowledge. However, current systems assume a single "best" description per concept, which fails to account for the many potentially useful ways a concept can be described. We present ACCoRD, an end-to-end system tackling the novel task of generating sets of descriptions of scientific concepts. Our system takes advantage of the myriad ways a concept is mentioned across the scientific literature to produce distinct, diverse descriptions of target scientific concepts in terms of different reference concepts. To support research on the task, we release an expert-annotated resource, the ACCoRD corpus, which includes 1,275 labeled contexts and 1,787 hand-authored concept descriptions. We conduct a user study demonstrating that (1) users prefer descriptions produced by our end-to-end system, and (2) users prefer multiple descriptions to a single "best" description.

Mandar Joshi, Terra Blevins, Mike Lewis et al. · uw

Creating labeled natural language training data is expensive and requires significant human effort. We mine input output examples from large corpora using a supervised mining function trained using a small seed set of only 100 examples. The mining consists of two stages -- (1) a biencoder-based recall-oriented dense search which pairs inputs with potential outputs, and (2) a crossencoder-based filter which re-ranks the output of the biencoder stage for better precision. Unlike model-generated data augmentation, our method mines naturally occurring high-quality input output pairs to mimic the style of the seed set for multiple tasks. On SQuAD-style reading comprehension, augmenting the seed set with the mined data results in an improvement of 13 F1 over a BART-large baseline fine-tuned only on the seed set. Likewise, we see improvements of 1.46 ROUGE-L on Xsum abstractive summarization.

Joyce Zhou, Elena Glassman, Daniel S. Weld · harvard, uw

Scientists and science journalists, among others, often need to make sense of a large number of papers and how they compare with each other in scope, focus, findings, or any other important factors. However, with a large corpus of papers, it's cognitively demanding to pairwise compare and contrast them all with each other. Fully automating this review process would be infeasible, because it often requires domain-specific knowledge, as well as understanding what the context and motivations for the review are. While there are existing tools to help with the process of organizing and annotating papers for literature reviews, at the core they still rely on people to serially read through papers and manually make sense of relevant information. We present AVTALER, which combines peoples' unique skills, contextual awareness, and knowledge, together with the strength of automation. Given a set of comparable text excerpts from a paper corpus, it supports users in sensemaking and contrasting paper attributes by interactively aligning text excerpts in a table so that comparable details are presented in a shared column. AVTALER is based on a core alignment algorithm that makes use of modern NLP tools. Furthermore, AVTALER is a mixed-initiative system: users can interactively give the system constraints which are integrated into the alignment construction process.

Marissa Radensky, Dustin Burson, Rajya Bhaiya et al. · uw

An important goal in the field of human-AI interaction is to help users more appropriately trust AI systems' decisions. A situation in which the user may particularly benefit from more appropriate trust is when the AI receives anomalous input or provides anomalous output. To the best of our knowledge, this is the first work towards understanding how anomaly alerts may contribute to appropriate trust of AI. In a formative mixed-methods study with 4 radiologists and 4 other physicians, we explore how AI alerts for anomalous input, very high and low confidence, and anomalous saliency-map explanations affect users' experience with mockups of an AI clinical decision support system (CDSS) for evaluating chest x-rays for pneumonia. We find evidence suggesting that the four anomaly alerts are desired by non-radiologists, and the high-confidence alerts are desired by both radiologists and non-radiologists. In a follow-up user study, we investigate how high- and low-confidence alerts affect the accuracy and thus appropriate trust of 33 radiologists working with AI CDSS mockups. We observe that these alerts do not improve users' accuracy or experience and discuss potential reasons why.

Anirudh Ajith, Amanpreet Singh, Jay DeYoung et al.

Can AI systems trained on the scientific record up to a fixed point in time forecast the scientific advances that follow? Such a capability could help researchers identify collaborators and impactful research directions, and anticipate which problems and methods will become central next. We introduce PreScience -- a scientific forecasting benchmark that decomposes the research process into four interdependent generative tasks: collaborator prediction, prior work selection, contribution generation, and impact prediction. PreScience is a carefully curated dataset of 98K recent AI-related research papers, featuring disambiguated author identities, temporally aligned scholarly metadata, and a structured graph of companion author publication histories and citations spanning 502K total papers. We develop baselines and evaluations for each task, including LACERScore, a novel LLM-based measure of contribution similarity that outperforms previous metrics and approximates inter-annotator agreement. We find substantial headroom remains in each task -- e.g. in contribution generation, frontier LLMs achieve only moderate similarity to the ground-truth (GPT-5, averages 5.6 on a 1-10 scale). When composed into a 12-month end-to-end simulation of scientific production, the resulting synthetic corpus is systematically less diverse and less novel than human-authored research from the same period.

Amanpreet Singh, Joseph Chee Chang, Chloe Anastasiades et al. · allen-ai

Retrieval-augmented generation is increasingly effective in answering scientific questions from literature, but many state-of-the-art systems are expensive and closed-source. We introduce Ai2 Scholar QA, a free online scientific question answering application. To facilitate research, we make our entire pipeline public: as a customizable open-source Python package and interactive web app, along with paper indexes accessible through public APIs and downloadable datasets. We describe our system in detail and present experiments analyzing its key design decisions. In an evaluation on a recent scientific QA benchmark, we find that Ai2 Scholar QA outperforms competing systems.

Pao Siangliulue, Jonathan Bragg, Doug Downey et al.

As AI agents become increasingly capable of complex knowledge tasks, the lack of context limits their capability to proactively reason about a user's latent needs throughout a long evolving project. In scientific research, many researchers still manually query a deep research system and compress their rich project contexts into short, targeted queries. Further, a deep research system produces exhaustive reports, making it difficult to identify concrete actions. To explore the opportunities of research assistants that are proactive throughout a research project, we conducted several studies (N=42) with a technology probe and an iterative prototype. The latest iteration of our system, Omakase, is a research assistant that monitors a user's project documents to infer timely queries to a deep research system. Omakase then distills long reports into suggestions contextualized to their evolving projects. Our evaluations showed that participants found the generated queries to be useful and timely, and rated Omakase's suggestions as significantly more actionable than the original reports.

Peter Jansen, Peter Clark, Doug Downey et al.

Contemporary automated scientific discovery has focused on agents for generating scientific experiments, while systems that perform higher-level scientific activities such as theory building remain underexplored. In this work, we formulate the problem of synthesizing theories consisting of qualitative and quantitative laws from large corpora of scientific literature. We study theory generation at scale, using 13.7k source papers to synthesize 2.9k theories, examining how generation using literature-grounding versus parametric knowledge, and accuracy-focused versus novelty-focused generation objectives change theory properties. Our experiments show that, compared to using parametric LLM memory for generation, our literature-supported method creates theories that are significantly better at both matching existing evidence and at predicting future results from 4.6k subsequently-written papers

Zejiang Shen, Kyle Lo, Lucy Lu Wang et al.

Accurately extracting structured content from PDFs is a critical first step for NLP over scientific papers. Recent work has improved extraction accuracy by incorporating elementary layout information, e.g., each token's 2D position on the page, into language model pretraining. We introduce new methods that explicitly model VIsual LAyout (VILA) groups, i.e., text lines or text blocks, to further improve performance. In our I-VILA approach, we show that simply inserting special tokens denoting layout group boundaries into model inputs can lead to a 1.9% Macro F1 improvement in token classification. In the H-VILA approach, we show that hierarchical encoding of layout-groups can result in up-to 47% inference time reduction with less than 0.8% Macro F1 loss. Unlike prior layout-aware approaches, our methods do not require expensive additional pretraining, only fine-tuning, which we show can reduce training cost by up to 95%. Experiments are conducted on a newly curated evaluation suite, S2-VLUE, that unifies existing automatically-labeled datasets and includes a new dataset of manual annotations covering diverse papers from 19 scientific disciplines. Pre-trained weights, benchmark datasets, and source code are available at https://github.com/allenai/VILA.

Daniel King, Doug Downey, Daniel S. Weld

Identification of new concepts in scientific literature can help power faceted search, scientific trend analysis, knowledge-base construction, and more, but current methods are lacking. Manual identification cannot keep up with the torrent of new publications, while the precision of existing automatic techniques is too low for many applications. We present an unsupervised concept extraction method for scientific literature that achieves much higher precision than previous work. Our approach relies on a simple but novel intuition: each scientific concept is likely to be introduced or popularized by a single paper that is disproportionately cited by subsequent papers mentioning the concept. From a corpus of computer science papers on arXiv, we find that our method achieves a Precision@1000 of 99%, compared to 86% for prior work, and a substantially better precision-yield trade-off across the top 15,000 extractions. To stimulate research in this area, we release our code and data (https://github.com/allenai/ForeCite).

Isabel Cachola, Kyle Lo, Arman Cohan et al.

We introduce TLDR generation, a new form of extreme summarization, for scientific papers. TLDR generation involves high source compression and requires expert background knowledge and understanding of complex domain-specific language. To facilitate study on this task, we introduce SciTLDR, a new multi-target dataset of 5.4K TLDRs over 3.2K papers. SciTLDR contains both author-written and expert-derived TLDRs, where the latter are collected using a novel annotation protocol that produces high-quality summaries while minimizing annotation burden. We propose CATTS, a simple yet effective learning strategy for generating TLDRs that exploits titles as an auxiliary training signal. CATTS improves upon strong baselines under both automated metrics and human evaluations. Data and code are publicly available at https://github.com/allenai/scitldr.

Mandar Joshi, Eunsol Choi, Daniel S. Weld et al.

We present TriviaQA, a challenging reading comprehension dataset containing over 650K question-answer-evidence triples. TriviaQA includes 95K question-answer pairs authored by trivia enthusiasts and independently gathered evidence documents, six per question on average, that provide high quality distant supervision for answering the questions. We show that, in comparison to other recently introduced large-scale datasets, TriviaQA (1) has relatively complex, compositional questions, (2) has considerable syntactic and lexical variability between questions and corresponding answer-evidence sentences, and (3) requires more cross sentence reasoning to find answers. We also present two baseline algorithms: a feature-based classifier and a state-of-the-art neural network, that performs well on SQuAD reading comprehension. Neither approach comes close to human performance (23% and 40% vs. 80%), suggesting that TriviaQA is a challenging testbed that is worth significant future study. Data and code available at -- http://nlp.cs.washington.edu/triviaqa/

Madeleine Grunde-McLaughlin, Michelle S. Lam, Ranjay Krishna et al. · uw

LLM chains enable complex tasks by decomposing work into a sequence of subtasks. Similarly, the more established techniques of crowdsourcing workflows decompose complex tasks into smaller tasks for human crowdworkers. Chains address LLM errors analogously to the way crowdsourcing workflows address human error. To characterize opportunities for LLM chaining, we survey 107 papers across the crowdsourcing and chaining literature to construct a design space for chain development. The design space covers a designer's objectives and the tactics used to build workflows. We then surface strategies that mediate how workflows use tactics to achieve objectives. To explore how techniques from crowdsourcing may apply to chaining, we adapt crowdsourcing workflows to implement LLM chains across three case studies: creating a taxonomy, shortening text, and writing a short story. From the design space and our case studies, we identify takeaways for effective chain design and raise implications for future research and development.

Peter Jansen, Oyvind Tafjord, Marissa Radensky et al. · allen-ai

Despite the surge of interest in autonomous scientific discovery (ASD) of software artifacts (e.g., improved ML algorithms), current ASD systems face two key limitations: (1) they largely explore variants of existing codebases or similarly constrained design spaces, and (2) they produce large volumes of research artifacts (such as automatically generated papers and code) that are typically evaluated using conference-style paper review with limited evaluation of code. In this work we introduce CodeScientist, a novel ASD system that frames ideation and experiment construction as a form of genetic search jointly over combinations of research articles and codeblocks defining common actions in a domain (like prompting a language model). We use this paradigm to conduct hundreds of automated experiments on machine-generated ideas broadly in the domain of agents and virtual environments, with the system returning 19 discoveries, 6 of which were judged as being both at least minimally sound and incrementally novel after a multi-faceted evaluation beyond that typically conducted in prior work, including external (conference-style) review, code review, and replication attempts. Moreover, the discoveries span new tasks, agents, metrics, and data, suggesting a qualitative shift from benchmark optimization to broader discoveries.

Gagan Bansal, Jennifer Wortman Vaughan, Saleema Amershi et al. · microsoft-research

Remarkable advancements in modern generative foundation models have enabled the development of sophisticated and highly capable autonomous agents that can observe their environment, invoke tools, and communicate with other agents to solve problems. Although such agents can communicate with users through natural language, their complexity and wide-ranging failure modes present novel challenges for human-AI interaction. Building on prior research and informed by a communication grounding perspective, we contribute to the study of \emph{human-agent communication} by identifying and analyzing twelve key communication challenges that these systems pose. These include challenges in conveying information from the agent to the user, challenges in enabling the user to convey information to the agent, and overarching challenges that need to be considered across all human-agent communication. We illustrate each challenge through concrete examples and identify open directions of research. Our findings provide insights into critical gaps in human-agent communication research and serve as an urgent call for new design patterns, principles, and guidelines to support transparency and control in these systems.

Benjamin Newman, Yoonjoo Lee, Aakanksha Naik et al. · allen-ai, uw

When conducting literature reviews, scientists often create literature review tables - tables whose rows are publications and whose columns constitute a schema, a set of aspects used to compare and contrast the papers. Can we automatically generate these tables using language models (LMs)? In this work, we introduce a framework that leverages LMs to perform this task by decomposing it into separate schema and value generation steps. To enable experimentation, we address two main challenges: First, we overcome a lack of high-quality datasets to benchmark table generation by curating and releasing arxivDIGESTables, a new dataset of 2,228 literature review tables extracted from ArXiv papers that synthesize a total of 7,542 research papers. Second, to support scalable evaluation of model generations against human-authored reference tables, we develop DecontextEval, an automatic evaluation method that aligns elements of tables with the same underlying aspects despite differing surface forms. Given these tools, we evaluate LMs' abilities to reconstruct reference tables, finding this task benefits from additional context to ground the generation (e.g. table captions, in-text references). Finally, through a human evaluation study we find that even when LMs fail to fully reconstruct a reference table, their generated novel aspects can still be useful.

Simra Shahid, Marissa Radensky, Raymond Fok et al. · allen-ai, uw

Automated scientific idea generation systems have made remarkable progress, yet the automatic evaluation of idea novelty remains a critical and underexplored challenge. Manual evaluation of novelty through literature review is labor-intensive, prone to error due to subjectivity, and impractical at scale. To address these issues, we propose the Idea Novelty Checker, an LLM-based retrieval-augmented generation (RAG) framework that leverages a two-stage retrieve-then-rerank approach. The Idea Novelty Checker first collects a broad set of relevant papers using keyword and snippet-based retrieval, then refines this collection through embedding-based filtering followed by facet-based LLM re-ranking. It incorporates expert-labeled examples to guide the system in comparing papers for novelty evaluation and in generating literature-grounded reasoning. Our extensive experiments demonstrate that our novelty checker achieves approximately 13% higher agreement than existing approaches. Ablation studies further showcases the importance of the facet-based re-ranker in identifying the most relevant literature for novelty evaluation.

Jonathan Bragg, Mike D'Arcy, Nishant Balepur et al. · allen-ai

AI agents hold the potential to revolutionize scientific productivity by automating literature reviews, replicating experiments, analyzing data, and even proposing new directions of inquiry; indeed, there are now many such agents, ranging from general-purpose "deep research" systems to specialized science-specific agents, such as AI Scientist and AIGS. Rigorous evaluation of these agents is critical for progress. Yet existing benchmarks fall short on several fronts: they (1) fail to provide holistic, product-informed measures of real-world use cases such as science research; (2) lack reproducible agent tools necessary for a controlled comparison of core agentic capabilities; (3) do not account for confounding variables such as model cost and tool access; (4) do not provide standardized interfaces for quick agent prototyping and evaluation; and (5) lack comprehensive baseline agents necessary to identify true advances. In response, we define principles and tooling for more rigorously benchmarking agents. Using these, we present AstaBench, a suite that provides the first holistic measure of agentic ability to perform scientific research, comprising 2400+ problems spanning the entire scientific discovery process and multiple scientific domains, and including many problems inspired by actual user requests to deployed Asta agents. Our suite comes with the first scientific research environment with production-grade search tools that enable controlled, reproducible evaluation, better accounting for confounders. Alongside, we provide a comprehensive suite of nine science-optimized classes of Asta agents and numerous baselines. Our extensive evaluation of 57 agents across 22 agent classes reveals several interesting findings, most importantly that despite meaningful progress on certain individual aspects, AI remains far from solving the challenge of science research assistance.

Raymond Fok, Daniel S. Weld

The current literature on AI-advised decision making -- involving explainable AI systems advising human decision makers -- presents a series of inconclusive and confounding results. To synthesize these findings, we propose a simple theory that elucidates the frequent failure of AI explanations to engender appropriate reliance and complementary decision making performance. We argue explanations are only useful to the extent that they allow a human decision maker to verify the correctness of an AI's prediction, in contrast to other desiderata, e.g., interpretability or spelling out the AI's reasoning process. Prior studies find in many decision making contexts AI explanations do not facilitate such verification. Moreover, most tasks fundamentally do not allow easy verification, regardless of explanation method, limiting the potential benefit of any type of explanation. We also compare the objective of complementary performance with that of appropriate reliance, decomposing the latter into the notions of outcome-graded and strategy-graded reliance.

Marissa Radensky, Doug Downey, Kyle Lo et al.

Explanations are well-known to improve recommender systems' transparency. These explanations may be local, explaining an individual recommendation, or global, explaining the recommender model in general. Despite their widespread use, there has been little investigation into the relative benefits of these two approaches. Do they provide the same benefits to users, or do they serve different purposes? We conducted a 30-participant exploratory study and a 30-participant controlled user study with a research-paper recommender system to analyze how providing participants local, global, or both explanations influences user understanding of system behavior. Our results provide evidence suggesting that both explanations are more helpful than either alone for explaining how to improve recommendations, yet both appeared less helpful than global alone for efficiency in identifying false positives and negatives. However, we note that the two explanation approaches may be better compared in the context of a higher-stakes or more opaque domain.

Dan Lahav, Jon Saad Falcon, Bailey Kuehl et al.

Keeping track of scientific challenges, advances and emerging directions is a fundamental part of research. However, researchers face a flood of papers that hinders discovery of important knowledge. In biomedicine, this directly impacts human lives. To address this problem, we present a novel task of extraction and search of scientific challenges and directions, to facilitate rapid knowledge discovery. We construct and release an expert-annotated corpus of texts sampled from full-length papers, labeled with novel semantic categories that generalize across many types of challenges and directions. We focus on a large corpus of interdisciplinary work relating to the COVID-19 pandemic, ranging from biomedicine to areas such as AI and economics. We apply a model trained on our data to identify challenges and directions across the corpus and build a dedicated search engine. In experiments with 19 researchers and clinicians using our system, we outperform a popular scientific search engine in assisting knowledge discovery. Finally, we show that models trained on our resource generalize to the wider biomedical domain and to AI papers, highlighting its broad utility. We make our data, model and search engine publicly available. https://challenges.apps.allenai.org/

Quanze Chen, Daniel S. Weld, Amy X. Zhang

Human ratings have become a crucial resource for training and evaluating machine learning systems. However, traditional elicitation methods for absolute and comparative rating suffer from issues with consistency and often do not distinguish between uncertainty due to disagreement between annotators and ambiguity inherent to the item being rated. In this work, we present Goldilocks, a novel crowd rating elicitation technique for collecting calibrated scalar annotations that also distinguishes inherent ambiguity from inter-annotator disagreement. We introduce two main ideas: grounding absolute rating scales with examples and using a two-step bounding process to establish a range for an item's placement. We test our designs in three domains: judging toxicity of online comments, estimating satiety of food depicted in images, and estimating age based on portraits. We show that (1) Goldilocks can improve consistency in domains where interpretation of the scale is not universal, and that (2) representing items with ranges lets us simultaneously capture different sources of uncertainty leading to better estimates of pairwise relationship distributions.

Lucy Lu Wang, Isabel Cachola, Jonathan Bragg et al.

The majority of scientific papers are distributed in PDF, which pose challenges for accessibility, especially for blind and low vision (BLV) readers. We characterize the scope of this problem by assessing the accessibility of 11,397 PDFs published 2010--2019 sampled across various fields of study, finding that only 2.4% of these PDFs satisfy all of our defined accessibility criteria. We introduce the SciA11y system to offset some of the issues around inaccessibility. SciA11y incorporates several machine learning models to extract the content of scientific PDFs and render this content as accessible HTML, with added novel navigational features to support screen reader users. An intrinsic evaluation of extraction quality indicates that the majority of HTML renders (87%) produced by our system have no or only some readability issues. We perform a qualitative user study to understand the needs of BLV researchers when reading papers, and to assess whether the SciA11y system could address these needs. We summarize our user study findings into a set of five design recommendations for accessible scientific reader systems. User response to SciA11y was positive, with all users saying they would be likely to use the system in the future, and some stating that the system, if available, would become their primary workflow. We successfully produce HTML renders for over 12M papers, of which an open access subset of 1.5M are available for browsing at https://scia11y.org/

Daniel Khashabi, Gabriel Stanovsky, Jonathan Bragg et al.

While often assumed a gold standard, effective human evaluation of text generation remains an important, open area for research. We revisit this problem with a focus on producing consistent evaluations that are reproducible -- over time and across different populations. We study this goal in different stages of the human evaluation pipeline. In particular, we consider design choices for the annotation interface used to elicit human judgments and their impact on reproducibility. Furthermore, we develop an automated mechanism for maintaining annotator quality via a probabilistic model that detects and excludes noisy annotators. Putting these lessons together, we introduce GENIE: a system for running standardized human evaluations across different generation tasks. We instantiate GENIE with datasets representing four core challenges in text generation: machine translation, summarization, commonsense reasoning, and machine comprehension. For each task, GENIE offers a leaderboard that automatically crowdsources annotations for submissions, evaluating them along axes such as correctness, conciseness, and fluency. We have made the GENIE leaderboards publicly available, and have already ranked 50 submissions from 10 different research groups. We hope GENIE encourages further progress toward effective, standardized evaluations for text generation.

Tongshuang Wu, Marco Tulio Ribeiro, Jeffrey Heer et al.

While counterfactual examples are useful for analysis and training of NLP models, current generation methods either rely on manual labor to create very few counterfactuals, or only instantiate limited types of perturbations such as paraphrases or word substitutions. We present Polyjuice, a general-purpose counterfactual generator that allows for control over perturbation types and locations, trained by finetuning GPT-2 on multiple datasets of paired sentences. We show that Polyjuice produces diverse sets of realistic counterfactuals, which in turn are useful in various distinct applications: improving training and evaluation on three different tasks (with around 70% less annotation effort than manual generation), augmenting state-of-the-art explanation techniques, and supporting systematic counterfactual error analysis by revealing behaviors easily missed by human experts.

Dongyeop Kang, Andrew Head, Risham Sidhu et al.

The task of definition detection is important for scholarly papers, because papers often make use of technical terminology that may be unfamiliar to readers. Despite prior work on definition detection, current approaches are far from being accurate enough to use in real-world applications. In this paper, we first perform in-depth error analysis of the current best performing definition detection system and discover major causes of errors. Based on this analysis, we develop a new definition detection system, HEDDEx, that utilizes syntactic features, transformer encoders, and heuristic filters, and evaluate it on a standard sentence-level benchmark. Because current benchmarks evaluate randomly sampled sentences, we propose an alternative evaluation that assesses every sentence within a document. This allows for evaluating recall in addition to precision. HEDDEx outperforms the leading system on both the sentence-level and the document-level tasks, by 12.7 F1 points and 14.4 F1 points, respectively. We note that performance on the high-recall document-level task is much lower than in the standard evaluation approach, due to the necessity of incorporation of document structure as features. We discuss remaining challenges in document-level definition detection, ideas for improvements, and potential issues for the development of reading aid applications.

Andrew Head, Kyle Lo, Dongyeop Kang et al.

Despite the central importance of research papers to scientific progress, they can be difficult to read. Comprehension is often stymied when the information needed to understand a passage resides somewhere else: in another section, or in another paper. In this work, we envision how interfaces can bring definitions of technical terms and symbols to readers when and where they need them most. We introduce ScholarPhi, an augmented reading interface with four novel features: (1) tooltips that surface position-sensitive definitions from elsewhere in a paper, (2) a filter over the paper that "declutters" it to reveal how the term or symbol is used across the paper, (3) automatic equation diagrams that expose multiple definitions in parallel, and (4) an automatically generated glossary of important terms and symbols. A usability study showed that the tool helps researchers of all experience levels read papers. Furthermore, researchers were eager to have ScholarPhi's definitions available to support their everyday reading.

Gagan Bansal, Tongshuang Wu, Joyce Zhou et al.

Many researchers motivate explainable AI with studies showing that human-AI team performance on decision-making tasks improves when the AI explains its recommendations. However, prior studies observed improvements from explanations only when the AI, alone, outperformed both the human and the best team. Can explanations help lead to complementary performance, where team accuracy is higher than either the human or the AI working solo? We conduct mixed-method user studies on three datasets, where an AI with accuracy comparable to humans helps participants solve a task (explaining itself in some conditions). While we observed complementary improvements from AI augmentation, they were not increased by explanations. Rather, explanations increased the chance that humans will accept the AI's recommendation, regardless of its correctness. Our result poses new challenges for human-centered AI: Can we develop explanatory approaches that encourage appropriate trust in AI, and therefore help generate (or improve) complementary performance?

Tom Hope, Jason Portenoy, Kishore Vasan et al.

The COVID-19 pandemic has sparked unprecedented mobilization of scientists, generating a deluge of papers that makes it hard for researchers to keep track and explore new directions. Search engines are designed for targeted queries, not for discovery of connections across a corpus. In this paper, we present SciSight, a system for exploratory search of COVID-19 research integrating two key capabilities: first, exploring associations between biomedical facets automatically extracted from papers (e.g., genes, drugs, diseases, patient outcomes); second, combining textual and network information to search and visualize groups of researchers and their ties. SciSight has so far served over $15K$ users with over $42K$ page views and $13\%$ returns.

Benjamin Charles Germain Lee, Jaime Mears, Eileen Jakeway et al.

Chronicling America is a product of the National Digital Newspaper Program, a partnership between the Library of Congress and the National Endowment for the Humanities to digitize historic newspapers. Over 16 million pages of historic American newspapers have been digitized for Chronicling America to date, complete with high-resolution images and machine-readable METS/ALTO OCR. Of considerable interest to Chronicling America users is a semantified corpus, complete with extracted visual content and headlines. To accomplish this, we introduce a visual content recognition model trained on bounding box annotations of photographs, illustrations, maps, comics, and editorial cartoons collected as part of the Library of Congress's Beyond Words crowdsourcing initiative and augmented with additional annotations including those of headlines and advertisements. We describe our pipeline that utilizes this deep learning model to extract 7 classes of visual content: headlines, photographs, illustrations, maps, comics, editorial cartoons, and advertisements, complete with textual content such as captions derived from the METS/ALTO OCR, as well as image embeddings for fast image similarity querying. We report the results of running the pipeline on 16.3 million pages from the Chronicling America corpus and describe the resulting Newspaper Navigator dataset, the largest dataset of extracted visual content from historic newspapers ever produced. The Newspaper Navigator dataset, finetuned visual content recognition model, and all source code are placed in the public domain for unrestricted re-use.

Gagan Bansal, Besmira Nushi, Ece Kamar et al.

AI practitioners typically strive to develop the most accurate systems, making an implicit assumption that the AI system will function autonomously. However, in practice, AI systems often are used to provide advice to people in domains ranging from criminal justice and finance to healthcare. In such AI-advised decision making, humans and machines form a team, where the human is responsible for making final decisions. But is the most accurate AI the best teammate? We argue "No" -- predictable performance may be worth a slight sacrifice in AI accuracy. Instead, we argue that AI systems should be trained in a human-centered manner, directly optimized for team performance. We study this proposal for a specific type of human-AI teaming, where the human overseer chooses to either accept the AI recommendation or solve the task themselves. To optimize the team performance for this setting we maximize the team's expected utility, expressed in terms of the quality of the final decision, cost of verifying, and individual accuracies of people and machines. Our experiments with linear and non-linear models on real-world, high-stakes datasets show that the most accuracy AI may not lead to highest team performance and show the benefit of modeling teamwork during training through improvements in expected team utility across datasets, considering parameters such as human skill and the cost of mistakes. We discuss the shortcoming of current optimization approaches beyond well-studied loss functions such as log-loss, and encourage future work on AI optimization problems motivated by human-AI collaboration.

Lucy Lu Wang, Kyle Lo, Yoganand Chandrasekhar et al.

The COVID-19 Open Research Dataset (CORD-19) is a growing resource of scientific papers on COVID-19 and related historical coronavirus research. CORD-19 is designed to facilitate the development of text mining and information retrieval systems over its rich collection of metadata and structured full text papers. Since its release, CORD-19 has been downloaded over 200K times and has served as the basis of many COVID-19 text mining and discovery systems. In this article, we describe the mechanics of dataset construction, highlighting challenges and key design decisions, provide an overview of how CORD-19 has been used, and describe several shared tasks built around the dataset. We hope this resource will continue to bring together the computing community, biomedical experts, and policy makers in the search for effective treatments and management policies for COVID-19.

Arman Cohan, Sergey Feldman, Iz Beltagy et al.

Representation learning is a critical ingredient for natural language processing systems. Recent Transformer language models like BERT learn powerful textual representations, but these models are targeted towards token- and sentence-level training objectives and do not leverage information on inter-document relatedness, which limits their document-level representation power. For applications on scientific documents, such as classification and recommendation, the embeddings power strong performance on end tasks. We propose SPECTER, a new method to generate document-level embedding of scientific documents based on pretraining a Transformer language model on a powerful signal of document-level relatedness: the citation graph. Unlike existing pretrained language models, SPECTER can be easily applied to downstream applications without task-specific fine-tuning. Additionally, to encourage further research on document-level models, we introduce SciDocs, a new evaluation benchmark consisting of seven document-level tasks ranging from citation prediction, to document classification and recommendation. We show that SPECTER outperforms a variety of competitive baselines on the benchmark.

Benjamin Charles Germain Lee, Doug Downey, Kyle Lo et al.

Research in human-centered AI has shown the benefits of systems that can explain their predictions. Methods that allow an AI to take advice from humans in response to explanations are similarly useful. While both capabilities are well-developed for transparent learning models (e.g., linear models and GA$^2$Ms), and recent techniques (e.g., LIME and SHAP) can generate explanations for opaque models, little attention has been given to advice methods for opaque models. This paper introduces LIMEADE, the first general framework that translates both positive and negative advice (expressed using high-level vocabulary such as that employed by post-hoc explanations) into an update to an arbitrary, underlying opaque model. We demonstrate the generality of our approach with case studies on seventy real-world models across two broad domains: image classification and text recommendation. We show our method improves accuracy compared to a rigorous baseline on the image classification domains. For the text modality, we apply our framework to a neural recommender system for scientific papers on a public website; our user study shows that our framework leads to significantly higher perceived user control, trust, and satisfaction.

Arman Cohan, Iz Beltagy, Daniel King et al.

As a step toward better document-level understanding, we explore classification of a sequence of sentences into their corresponding categories, a task that requires understanding sentences in context of the document. Recent successful models for this task have used hierarchical models to contextualize sentence representations, and Conditional Random Fields (CRFs) to incorporate dependencies between subsequent labels. In this work, we show that pretrained language models, BERT (Devlin et al., 2018) in particular, can be used for this task to capture contextual dependencies without the need for hierarchical encoding nor a CRF. Specifically, we construct a joint sentence representation that allows BERT Transformer layers to directly utilize contextual information from all words in all sentences. Our approach achieves state-of-the-art results on four datasets, including a new dataset of structured scientific abstracts.

Mandar Joshi, Omer Levy, Daniel S. Weld et al.

We apply BERT to coreference resolution, achieving strong improvements on the OntoNotes (+3.9 F1) and GAP (+11.5 F1) benchmarks. A qualitative analysis of model predictions indicates that, compared to ELMo and BERT-base, BERT-large is particularly better at distinguishing between related but distinct entities (e.g., President and CEO). However, there is still room for improvement in modeling document-level context, conversations, and mention paraphrasing. Our code and models are publicly available.

Mandar Joshi, Danqi Chen, Yinhan Liu et al.

We present SpanBERT, a pre-training method that is designed to better represent and predict spans of text. Our approach extends BERT by (1) masking contiguous random spans, rather than random tokens, and (2) training the span boundary representations to predict the entire content of the masked span, without relying on the individual token representations within it. SpanBERT consistently outperforms BERT and our better-tuned baselines, with substantial gains on span selection tasks such as question answering and coreference resolution. In particular, with the same training data and model size as BERT-large, our single model obtains 94.6% and 88.7% F1 on SQuAD 1.1 and 2.0, respectively. We also achieve a new state of the art on the OntoNotes coreference resolution task (79.6\% F1), strong performance on the TACRED relation extraction benchmark, and even show gains on GLUE.

Quanze Chen, Jonathan Bragg, Lydia B. Chilton et al.

Traditional approaches for ensuring high quality crowdwork have failed to achieve high-accuracy on difficult problems. Aggregating redundant answers often fails on the hardest problems when the majority is confused. Argumentation has been shown to be effective in mitigating these drawbacks. However, existing argumentation systems only support limited interactions and show workers general justifications, not context-specific arguments targeted to their reasoning. This paper presents Cicero, a new workflow that improves crowd accuracy on difficult tasks by engaging workers in multi-turn, contextual discussions through real-time, synchronous argumentation. Our experiments show that compared to previous argumentation systems which only improve the average individual worker accuracy by 6.8 percentage points on the Relation Extraction domain, our workflow achieves 16.7 percentage point improvement. Furthermore, previous argumentation approaches don't apply to tasks with many possible answers; in contrast, Cicero works well in these cases, raising accuracy from 66.7% to 98.8% on the Codenames domain.

Mandar Joshi, Eunsol Choi, Omer Levy et al.

Reasoning about implied relationships (e.g., paraphrastic, common sense, encyclopedic) between pairs of words is crucial for many cross-sentence inference problems. This paper proposes new methods for learning and using embeddings of word pairs that implicitly represent background knowledge about such relationships. Our pairwise embeddings are computed as a compositional function on word representations, which is learned by maximizing the pointwise mutual information (PMI) with the contexts in which the two words co-occur. We add these representations to the cross-sentence attention layer of existing inference models (e.g. BiDAF for QA, ESIM for NLI), instead of extending or replacing existing word embeddings. Experiments show a gain of 2.7% on the recently released SQuAD2.0 and 1.3% on MultiNLI. Our representations also aid in better generalization with gains of around 6-7% on adversarial SQuAD datasets, and 8.8% on the adversarial entailment test set by Glockner et al. (2018).

James Ferguson, Colin Lockard, Daniel S. Weld et al.

Supervised event extraction systems are limited in their accuracy due to the lack of available training data. We present a method for self-training event extraction systems by bootstrapping additional training data. This is done by taking advantage of the occurrence of multiple mentions of the same event instances across newswire articles from multiple sources. If our system can make a highconfidence extraction of some mentions in such a cluster, it can then acquire diverse training examples by adding the other mentions as well. Our experiments show significant performance improvements on multiple event extractors over ACE 2005 and TAC-KBP 2015 datasets.

Ziyu Yao, Daniel S. Weld, Wei-Peng Chen et al.

Stack Overflow (SO) has been a great source of natural language questions and their code solutions (i.e., question-code pairs), which are critical for many tasks including code retrieval and annotation. In most existing research, question-code pairs were collected heuristically and tend to have low quality. In this paper, we investigate a new problem of systematically mining question-code pairs from Stack Overflow (in contrast to heuristically collecting them). It is formulated as predicting whether or not a code snippet is a standalone solution to a question. We propose a novel Bi-View Hierarchical Neural Network which can capture both the programming content and the textual context of a code snippet (i.e., two views) to make a prediction. On two manually annotated datasets in Python and SQL domain, our framework substantially outperforms heuristic methods with at least 15% higher F1 and accuracy. Furthermore, we present StaQC (Stack Overflow Question-Code pairs), the largest dataset to date of ~148K Python and ~120K SQL question-code pairs, automatically mined from SO using our framework. Under various case studies, we demonstrate that StaQC can greatly help develop data-hungry models for associating natural language with programming language.

Daniel S. Weld, Gagan Bansal

Since Artificial Intelligence (AI) software uses techniques like deep lookahead search and stochastic optimization of huge neural networks to fit mammoth datasets, it often results in complex behavior that is difficult for people to understand. Yet organizations are deploying AI algorithms in many mission-critical settings. To trust their behavior, we must make AI intelligible, either by using inherently interpretable models or by developing new methods for explaining and controlling otherwise overwhelmingly complex decisions using local approximation, vocabulary alignment, and interactive explanation. This paper argues that intelligibility is essential, surveys recent work on building such systems, and highlights key directions for research.

Christopher H. Lin, Mausam, Daniel S. Weld

We present POAPS, a novel planning system for defining Partially Observable Markov Decision Processes (POMDPs) that abstracts away from POMDP details for the benefit of non-expert practitioners. POAPS includes an expressive adaptive programming language based on Lisp that has constructs for choice points that can be dynamically optimized. Non-experts can use our language to write adaptive programs that have partially observable components without needing to specify belief/hidden states or reason about probabilities. POAPS is also a compiler that defines and performs the transformation of any program written in our language into a POMDP with control knowledge. We demonstrate the generality and power of POAPS in the rapidly growing domain of human computation by describing its expressiveness and simplicity by writing several POAPS programs for common crowdsourcing tasks.