Large Language Models (LLMs) have issues with document question answering (QA) in situations where the document is unable to fit in the small context length of an LLM. To overcome this issue, most existing works focus on retrieving the relevant context from the document, representing them as plain text. However, documents such as PDFs, web pages, and presentations are naturally structured with different pages, tables, sections, and so on. Representing such structured documents as plain text is incongruous with the user’s mental model of these documents with rich structure. When a system has to query the document for context, this incongruity is brought to the fore, and seemingly trivial questions can trip up the QA system. To bridge this fundamental gap in handling structured documents, we propose an approach called PDFTriage that enables models to retrieve the context based on either structure or content. Our experiments demonstrate the effectiveness of the proposed PDFTriage-augmented models across several classes of questions where existing retrieval-augmented LLMs fail. To facilitate further research on this fundamental problem, we release our benchmark dataset consisting of 900+ human-generated questions over 80 structured documents from 10 different categories of question types for document QA. Our code and datasets will be released soon on Github.

Modern instruction-tuned models have become highly capable in text generation tasks such as summarization, and are expected to be released at a steady pace. In practice one may now wish to choose confidently, but with minimal effort, the best performing summarization model when applied to a new domain or purpose. In this work, we empirically investigate the test sample size necessary to select a preferred model in the context of news summarization. Empirical results reveal that comparative evaluation converges quickly for both automatic and human evaluation, with clear preferences for a system emerging from under 100 examples. The human preference data allows us to quantify how well automatic scores can reproduce preference rankings across a variety of downstream summarization tasks. We find that, while automatic metrics are stable at smaller sample sizes, only some automatic metrics are able to moderately predict model win rates according to human preference.

To achieve state-of-the-art performance, one still needs to train NER models on large-scale, high-quality annotated data, an asset that is both costly and time-intensive to accumulate. In contrast, real-world applications often resort to massive low-quality labeled data through non-expert annotators via crowdsourcing and external knowledge bases via distant supervision as a cost-effective alternative. However, these annotation methods result in noisy labels, which in turn lead to a notable decline in performance. Hence, we propose to denoise the noisy NER data with guidance from a small set of clean instances. Along with the main NER model we train a discriminator model and use its outputs to recalibrate the sample weights. The discriminator is capable of detecting both span and category errors with different discriminative prompts. Results on public crowdsourcing and distant supervision datasets show that the proposed method can consistently improve performance with a small guidance set.

The Portable Document Format (PDF) is a popular format for distributing digital documents. Datasets on PDF reading behaviors and interactions remain limited due to the challenges of instrumenting PDF readers for these data collection tasks. We present ATLAS, a data collection tool designed to better support researchers in collecting rich PDF-centric datasets from users. ATLAS supports researchers in programmatically creating a user interface for data collection that is ready to share with annotators. It includes a toolkit and an extensible schema to easily customize the data collection tasks for a variety of purposes, allowing collection of PDF annotations (e.g., highlights, drawings) as well as reading behavior analytics (e.g., page scroll, text selections). We open-source ATLAS1 to support future research efforts and review use cases of ATLAS that showcase our system’s broad applicability.

In real-world scenarios, labeled samples for dialogue summarization are usually limited (i.e., few-shot) due to high annotation costs for high-quality dialogue summaries. To efficiently learn from few-shot samples, previous works have utilized massive annotated data from other downstream tasks and then performed prompt transfer in prompt tuning so as to enable cross-task knowledge transfer. However, existing general-purpose prompt transfer techniques lack consideration for dialogue-specific information. In this paper, we focus on improving the prompt transfer from dialogue state tracking to dialogue summarization and propose Skeleton-Assisted Prompt Transfer (SAPT), which leverages skeleton generation as extra supervision that functions as a medium connecting the distinct source and target task and resulting in the model’s better consumption of dialogue state information. To automatically extract dialogue skeletons as supervised training data for skeleton generation, we design a novel approach with perturbation-based probes requiring neither annotation effort nor domain knowledge. Training the model on such skeletons can also help preserve model capability during prompt transfer. Our method significantly outperforms existing baselines. In-depth analyses demonstrate the effectiveness of our method in facilitating cross-task knowledge transfer in few-shot dialogue summarization.

In the task of entity description generation, given a context and a specified entity, a model must describe that entity correctly and in a contextually-relevant way. In this task, as well as broader language generation tasks, the generation of a nonfactual description (factual error) versus an incongruous description (contextual error) is fundamentally different, yet often conflated. We develop an evaluation paradigm that enables us to disentangle these two types of errors in naturally occurring textual contexts. We find that factuality and congruity are often at odds, and that models specifically struggle with accurate descriptions of entities that are less familiar to people. This shortcoming of language models raises concerns around the trustworthiness of such models, since factual errors on less well-known entities are exactly those that a human reader will not recognize.

Machine Learning models have lower accuracy when tested on out-of-domain data. Developing models that perform well on several domains or can be quickly adapted to a new domain is an important research area. Domain, however, is a vague term, that can refer to any aspect of data such as language, genre, source and structure. We consider a very homogeneous source of data, specifically sentences from news articles from the same newspaper in English, and collect a dataset of such “in-domain” sentences annotated with named entities. We find that even in such a homogeneous domain, the performance of named entity recognition models varies significantly across news topics. Selection of diverse data, as we demonstrate, is crucial even in a seemingly homogeneous domain.

Large-scale pre-training is widely used in recent document understanding tasks. During deployment, one may expect that models should trigger a conservative fallback policy when encountering out-of-distribution (OOD) samples, which highlights the importance of OOD detection. However, most existing OOD detection methods focus on single-modal inputs such as images or texts. While documents are multi-modal in nature, it is underexplored if and how multi-modal information in documents can be exploited for OOD detection. In this work, we first provide a systematic and in-depth analysis on OOD detection for document understanding models. We study the effects of model modality, pre-training, and fine-tuning across various types of OOD inputs. In particular, we find that spatial information is critical for document OOD detection. To better exploit spatial information, we propose a spatial-aware adapter, which serves as a parameter-efficient add-on module to adapt transformer-based language models to the document domain. Extensive experiments show that adding the spatial-aware adapter significantly improves the OOD detection performance compared to directly using the language model and achieves superior performance compared to competitive baselines.

Visual text evokes an image in a person’s mind, while non-visual text fails to do so. A method to automatically detect visualness in text will enable text-to-image retrieval and generation models to augment text with relevant images. This is particularly challenging with long-form text as text-to-image generation and retrieval models are often triggered for text that is designed to be explicitly visual in nature, whereas long-form text could contain many non-visual sentences. To this end, we curate a dataset of 3,620 English sentences and their visualness scores provided by multiple human annotators. We also propose a fine-tuning strategy that adapts large vision-language models like CLIP by modifying the model’s contrastive learning objective to map text identified as non-visual to a common NULL image while matching visual text to their corresponding images in the document. We evaluate the proposed approach on its ability to (i) classify visual and non-visual text accurately, and (ii) attend over words that are identified as visual in psycholinguistic studies. Empirical evaluation indicates that our approach performs better than several heuristics and baseline models for the proposed task. Furthermore, to highlight the importance of modeling the visualness of text, we conduct qualitative analyses of text-to-image generation systems like DALL-E.

Good figure captions help paper readers understand complex scientific figures. Unfortunately, even published papers often have poorly written captions. Automatic caption generation could aid paper writers by providing good starting captions that can be refined for better quality. Prior work often treated figure caption generation as a vision-to-language task. In this paper, we show that it can be more effectively tackled as a text summarization task in scientific documents. We fine-tuned PEGASUS, a pre-trained abstractive summarization model, to specifically summarize figure-referencing paragraphs (e.g., “Figure 3 shows...”) into figure captions. Experiments on large-scale arXiv figures show that our method outperforms prior vision methods in both automatic and human evaluations. We further conducted an in-depth investigation focused on two key challenges: (i) the common presence of low-quality author-written captions and (ii) the lack of clear standards for good captions. Our code and data are available at: https://github.com/Crowd-AI-Lab/Generating-Figure-Captions-as-a-Text-Summarization-Task.

Existing literature does not give much guidance on how to build the best possible multi-domain summarization model from existing components. We present an extensive evaluation of popular pre-trained models on a wide range of datasets to inform the selection of both the model and the training data for robust summarization across several domains. We find that fine-tuned BART performs better than T5 and PEGASUS, both on in-domain and out-of-domain data, regardless of the dataset used for fine-tuning. While BART has the best performance, it does vary considerably across domains. A multi-domain summarizer that works well for all domains can be built by simply fine-tuning on diverse domains. It even performs better than an in-domain summarizer, even when using fewer total training examples. While the success of such a multi-domain summarization model is clear through automatic evaluation, by conducting a human evaluation, we find that there are variations that can not be captured by any of the automatic evaluation metrics and thus not reflected in standard leaderboards. Furthermore, we find that conducting reliable human evaluation can be complex as well. Even experienced summarization researchers can be inconsistent with one another in their assessment of the quality of a summary, and also with themselves when re-annotating the same summary. The findings of our study are two-fold. First, BART fine-tuned on heterogeneous domains is a great multi-domain summarizer for practical purposes. At the same time, we need to re-examine not just automatic evaluation metrics but also human evaluation methods to responsibly measure progress in summarization.

We introduce DocTime - a novel temporal dependency graph (TDG) parser that takes as input a text document and produces a temporal dependency graph. It outperforms previous BERT-based solutions by a relative 4-8% on three datasets from modeling the problem as a graph network with path-prediction loss to incorporate longer range dependencies. This work also demonstrates how the TDG graph can be used to improve the downstream tasks of temporal questions answering and NLI by a relative 4-10% with a new framework that incorporates the temporal dependency graph into the self-attention layer of Transformer models (Time-transformer). Finally, we develop and evaluate on a new temporal dependency graph dataset for the domain of contractual documents, which has not been previously explored in this setting.

Document images are a ubiquitous source of data where the text is organized in a complex hierarchical structure ranging from fine granularity (e.g., words), medium granularity (e.g., regions such as paragraphs or figures), to coarse granularity (e.g., the whole page). The spatial hierarchical relationships between content at different levels of granularity are crucial for document image understanding tasks. Existing methods learn features from either word-level or region-level but fail to consider both simultaneously. Word-level models are restricted by the fact that they originate from pure-text language models, which only encode the word-level context. In contrast, region-level models attempt to encode regions corresponding to paragraphs or text blocks into a single embedding, but they perform worse with additional word-level features. To deal with these issues, we propose MGDoc, a new multi-modal multi-granular pre-training framework that encodes page-level, region-level, and word-level information at the same time. MGDoc uses a unified text-visual encoder to obtain multi-modal features across different granularities, which makes it possible to project the multi-granular features into the same hyperspace. To model the region-word correlation, we design a cross-granular attention mechanism and specific pre-training tasks for our model to reinforce the model of learning the hierarchy between regions and words. Experiments demonstrate that our proposed model can learn better features that perform well across granularities and lead to improvements in downstream tasks.

We present a comprehensive study of sparse attention patterns in Transformer models. We first question the need for pre-training with sparse attention and present experiments showing that an efficient fine-tuning only approach yields a slightly worse but still competitive model. Then we compare the widely used local attention pattern and the less-well-studied global attention pattern, demonstrating that global patterns have several unique advantages. We also demonstrate that a flexible approach to attention, with different patterns across different layers of the model, is beneficial for some tasks. Drawing on this insight, we propose a novel Adaptive Axis Attention method, which learns—during fine-tuning—different attention patterns for each Transformer layer depending on the downstream task. Rather than choosing a fixed attention pattern, the adaptive axis attention method identifies important tokens—for each task and model layer—and focuses attention on those. It does not require pre-training to accommodate the sparse patterns and demonstrates competitive and sometimes better performance against fixed sparse attention patterns that require resource-intensive pre-training.

Many standard tasks in NLP (e.g., Named Entity Recognition, Part-of-Speech tagging, and Semantic Role Labeling) are naturally framed as sequence tagging problems. However, there has been comparatively little work on interpretability methods for sequence tagging models. In this paper, we extend influence functions — which aim to trace predictions back to the training points that informed them — to sequence tagging tasks. We define the influence of a training instance segment as the effect that perturbing the labels within this segment has on a test segment level prediction. We provide an efficient approximation to compute this, and show that it tracks with the “true” segment influence (measured empirically). We show the practical utility of segment influence by using the method to identify noisy annotations in NER corpora.

Supervised training of existing deep learning models for sequence labeling relies on large scale labeled datasets. Such datasets are generally created with crowd-source labeling. However, crowd-source labeling for tasks of sequence labeling can be expensive and time-consuming. Further, crowd-source labeling by external annotators may not be appropriate for data that contains user private information. Considering the above limitations of crowd-source labeling, we study interactive sequence labeling that allows training directly with the user feedback, which alleviates the annotation cost and maintains the user privacy. We identify two bias, namely, context bias and feedback bias, by formulating interactive sequence labeling via a Structural Causal Model (SCM). To alleviate the context and feedback bias based on the SCM, we identify the frequent context tokens as confounders in the backdoor adjustment and further propose an entropy-based modulation that is inspired by information theory. entities more sample-efficiently. With extensive experiments, we validate that our approach can effectively alleviate the biases and our models can be efficiently learnt with the user feedback.

We provide a quantitative and qualitative analysis of self-repetition in the output of neural summarizers. We measure self-repetition as the number of n-grams of length four or longer that appear in multiple outputs of the same system. We analyze the behavior of three popular architectures (BART, T5, and Pegasus), fine-tuned on five datasets. In a regression analysis, we find that the three architectures have different propensities for repeating content across output summaries for inputs, with BART being particularly prone to self-repetition. Fine-tuning on more abstractive data, and on data featuring formulaic language is associated with a higher rate of self-repetition. In qualitative analysis, we find systems produce artefacts such as ads and disclaimers unrelated to the content being summarized, as well as formulaic phrases common in the fine-tuning domain. Our approach to corpus-level analysis of self-repetition may help practitioners clean up training data for summarizers and ultimately support methods for minimizing the amount of self-repetition.

Providing the right amount of explanation in an employment interview can help the interviewee effectively communicate their skills and experience to the interviewer and convince the she/he is the right candidate for the job. This paper examines natural language processing (NLP) approaches, including word-based tokenization, lexicon-based representations, and pre-trained embeddings with deep learning models, for detecting the degree of explanation in a job interview response. These are exemplified in a study of 24 military veterans who are the focal group of this study, since they can experience unique challenges in job interviews due to the unique verbal communication style that is prevalent in the military. Military veterans conducted mock interviews with industry recruiters and data from these interviews were transcribed and analyzed. Results indicate that the feasibility of automated NLP methods for detecting the degree of explanation in an interview response. Features based on tokenizer analysis are the most effective in detecting under-explained responses (i.e., 0.29 F1-score), while lexicon-based methods depict the higher performance in detecting over-explanation (i.e., 0.51 F1-score). Findings from this work lay the foundation for the design of intelligent assistive technologies that can provide personalized learning pathways to job candidates, especially those belonging to sensitive or under-represented populations, and helping them succeed in employment job interviews, ultimately contributing to an inclusive workforce.

Keeping the performance of language technologies optimal as time passes is of great practical interest. We study temporal effects on model performance on downstream language tasks, establishing a nuanced terminology for such discussion and identifying factors essential to conduct a robust study. We present experiments for several tasks in English where the label correctness is not dependent on time and demonstrate the importance of distinguishing between temporal model deterioration and temporal domain adaptation for systems using pre-trained representations. We find that, depending on the task, temporal model deterioration is not necessarily a concern. Temporal domain adaptation, however, is beneficial in all cases, with better performance for a given time period possible when the system is trained on temporally more recent data. Therefore, we also examine the efficacy of two approaches for temporal domain adaptation without human annotations on new data. Self-labeling shows consistent improvement and notably, for named entity recognition, leads to better temporal adaptation than even human annotations.

Named entity recognition systems achieve remarkable performance on domains such as English news. It is natural to ask: What are these models actually learning to achieve this? Are they merely memorizing the names themselves? Or are they capable of interpreting the text and inferring the correct entity type from the linguistic context? We examine these questions by contrasting the performance of several variants of architectures for named entity recognition, with some provided only representations of the context as features. We experiment with GloVe-based BiLSTM-CRF as well as BERT. We find that context does influence predictions, but the main factor driving high performance is learning the named tokens themselves. Furthermore, we find that BERT is not always better at recognizing predictive contexts compared to a BiLSTM-CRF model. We enlist human annotators to evaluate the feasibility of inferring entity types from context alone and find that humans are also mostly unable to infer entity types for the majority of examples on which the context-only system made errors. However, there is room for improvement: A system should be able to recognize any named entity in a predictive context correctly and our experiments indicate that current systems may be improved by such capability. Our human study also revealed that systems and humans do not always learn the same contextual clues, and context-only systems are sometimes correct even when humans fail to recognize the entity type from the context. Finally, we find that one issue contributing to model errors is the use of “entangled” representations that encode both contextual and local token information into a single vector, which can obscure clues. Our results suggest that designing models that explicitly operate over representations of local inputs and context, respectively, may in some cases improve performance. In light of these and related findings, we highlight directions for future work.

The ability to quantify incivility online, in news and in congressional debates is of great interest to political scientists. Computational tools for detecting online incivility for English are now fairly accessible and potentially could be applied more broadly. We test the Jigsaw Perspective API for its ability to detect the degree of incivility on a corpus that we developed, consisting of manual annotations of civility in American news. We demonstrate that toxicity models, as exemplified by Perspective, are inadequate for the analysis of incivility in news. We carry out error analysis that points to the need to develop methods to remove spurious correlations between words often mentioned in the news, especially identity descriptors and incivility. Without such improvements, applying Perspective or similar models on news is likely to lead to wrong conclusions, that are not aligned with the human perception of incivility.

We introduce Trialstreamer, a living database of clinical trial reports. Here we mainly describe the evidence extraction component; this extracts from biomedical abstracts key pieces of information that clinicians need when appraising the literature, and also the relations between these. Specifically, the system extracts descriptions of trial participants, the treatments compared in each arm (the interventions), and which outcomes were measured. The system then attempts to infer which interventions were reported to work best by determining their relationship with identified trial outcome measures. In addition to summarizing individual trials, these extracted data elements allow automatic synthesis of results across many trials on the same topic. We apply the system at scale to all reports of randomized controlled trials indexed in MEDLINE, powering the automatic generation of evidence maps, which provide a global view of the efficacy of different interventions combining data from all relevant clinical trials on a topic. We make all code and models freely available alongside a demonstration of the web interface.

Word representations trained on text reproduce human implicit bias related to gender, race and age. Methods have been developed to remove such bias. Here, we present results that show that human stereotypes exist even for much more nuanced judgments such as personality, for a variety of person identities beyond the typically legally protected attributes and that these are similarly captured in word representations. Specifically, we collected human judgments about a person’s Big Five personality traits formed solely from information about the occupation, nationality or a common noun description of a hypothetical person. Analysis of the data reveals a large number of statistically significant stereotypes in people. We then demonstrate the bias captured in lexical representations is statistically significantly correlated with the documented human bias. Our results, showing bias for a large set of person descriptors for such nuanced traits put in doubt the feasibility of broadly and fairly applying debiasing methods and call for the development of new methods for auditing language technology systems and resources.

Modern NLP systems require high-quality annotated data. For specialized domains, expert annotations may be prohibitively expensive; the alternative is to rely on crowdsourcing to reduce costs at the risk of introducing noise. In this paper we demonstrate that directly modeling instance difficulty can be used to improve model performance and to route instances to appropriate annotators. Our difficulty prediction model combines two learned representations: a ‘universal’ encoder trained on out of domain data, and a task-specific encoder. Experiments on a complex biomedical information extraction task using expert and lay annotators show that: (i) simply excluding from the training data instances predicted to be difficult yields a small boost in performance; (ii) using difficulty scores to weight instances during training provides further, consistent gains; (iii) assigning instances predicted to be difficult to domain experts is an effective strategy for task routing. Further, our experiments confirm the expectation that for such domain-specific tasks expert annotations are of much higher quality and preferable to obtain if practical and that augmenting small amounts of expert data with a larger set of lay annotations leads to further improvements in model performance.

It has been established that the performance of speech recognition systems depends on multiple factors including the lexical content, speaker identity and dialect. Here we use three English datasets of acted emotion to demonstrate that emotional content also impacts the performance of commercial systems. On two of the corpora, emotion is a bigger contributor to recognition errors than speaker identity and on two, neutral speech is recognized considerably better than emotional speech. We further evaluate the commercial systems on spontaneous interactions that contain portions of emotional speech. We propose and validate on the acted datasets, a method that allows us to evaluate the overall impact of emotion on recognition even when manual transcripts are not available. Using this method, we show that emotion in natural spontaneous dialogue is a less prominent but still significant factor in recognition accuracy.

ROUGE is widely used to automatically evaluate summarization systems. However, ROUGE measures semantic overlap between a system summary and a human reference on word-string level, much at odds with the contemporary treatment of semantic meaning. Here we present a suite of experiments on using distributed representations for evaluating summarizers, both in reference-based and in reference-free setting. Our experimental results show that the max value over each dimension of the summary ELMo word embeddings is a good representation that results in high correlation with human ratings. Averaging the cosine similarity of all encoders we tested yields high correlation with manual scores in reference-free setting. The distributed representations outperform ROUGE in recent corpora for abstractive news summarization but are less good on test data used in past evaluations.

We show that plain ROUGE F1 scores are not ideal for comparing current neural systems which on average produce different lengths. This is due to a non-linear pattern between ROUGE F1 and summary length. To alleviate the effect of length during evaluation, we have proposed a new method which normalizes the ROUGE F1 scores of a system by that of a random system with same average output length. A pilot human evaluation has shown that humans prefer short summaries in terms of the verbosity of a summary but overall consider longer summaries to be of higher quality. While human evaluations are more expensive in time and resources, it is clear that normalization, such as the one we proposed for automatic evaluation, will make human evaluations more meaningful.

Standard paradigms for search do not work well in the medical context. Typical information needs, such as retrieving a full list of medical interventions for a given condition, or finding the reported efficacy of a particular treatment with respect to a specific outcome of interest cannot be straightforwardly posed in typical text-box search. Instead, we propose faceted-search in which a user specifies a condition and then can browse treatments and outcomes that have been evaluated. Choosing from these, they can access randomized control trials (RCTs) describing individual studies. Realizing such a view of the medical evidence requires information extraction techniques to identify the population, interventions, and outcome measures in an RCT. Patients, health practitioners, and biomedical librarians all stand to benefit from such innovation in search of medical evidence. We present an initial prototype of such an interface applied to pre-registered clinical studies. We also discuss pilot studies into the applicability of information extraction methods to allow for similar access to all published trial results.

In many NLP applications like search and information extraction for named entities, it is necessary to find all the mentions of a named entity, some of which appear as pronouns (she, his, etc.) or nominals (the professor, the German chancellor, etc.). It is therefore important that coreference resolution systems are able to link these different types of mentions to the correct entity name. We evaluate state-of-the-art coreference resolution systems for the task of resolving all mentions to named entities. Our analysis reveals that standard coreference metrics do not reflect adequately the requirements in this task: they do not penalize systems for not identifying any mentions by name to an entity and they reward systems even if systems find correctly mentions to the same entity but fail to link these to a proper name (she–the student–no name). We introduce new metrics for evaluating named entity coreference that address these discrepancies and show that for the comparisons of competitive systems, standard coreference evaluations could give misleading results for this task. We are, however, able to confirm that the state-of-the art system according to traditional evaluations also performs vastly better than other systems on the named entity coreference task.

Medical professionals search the published literature by specifying the type of patients, the medical intervention(s) and the outcome measure(s) of interest. In this paper we demonstrate how features encoding syntactic patterns improve the performance of state-of-the-art sequence tagging models (both neural and linear) for information extraction of these medically relevant categories. We present an analysis of the type of patterns exploited and of the semantic space induced for these, i.e., the distributed representations learned for identified multi-token patterns. We show that these learned representations differ substantially from those of the constituent unigrams, suggesting that the patterns capture contextual information that is otherwise lost.

We present a corpus of 5,000 richly annotated abstracts of medical articles describing clinical randomized controlled trials. Annotations include demarcations of text spans that describe the Patient population enrolled, the Interventions studied and to what they were Compared, and the Outcomes measured (the ‘PICO’ elements). These spans are further annotated at a more granular level, e.g., individual interventions within them are marked and mapped onto a structured medical vocabulary. We acquired annotations from a diverse set of workers with varying levels of expertise and cost. We describe our data collection process and the corpus itself in detail. We then outline a set of challenging NLP tasks that would aid searching of the medical literature and the practice of evidence-based medicine.

Practical summarization systems are expected to produce summaries of varying lengths, per user needs. While a couple of early summarization benchmarks tested systems across multiple summary lengths, this practice was mostly abandoned due to the assumed cost of producing reference summaries of multiple lengths. In this paper, we raise the research question of whether reference summaries of a single length can be used to reliably evaluate system summaries of multiple lengths. For that, we have analyzed a couple of datasets as a case study, using several variants of the ROUGE metric that are standard in summarization evaluation. Our findings indicate that the evaluation protocol in question is indeed competitive. This result paves the way to practically evaluating varying-length summaries with simple, possibly existing, summarization benchmarks.

We present a robust approach for detecting intrinsic sentence importance in news, by training on two corpora of document-summary pairs. When used for single-document summarization, our approach, combined with the “beginning of document” heuristic, outperforms a state-of-the-art summarizer and the beginning-of-article baseline in both automatic and manual evaluations. These results represent an important advance because in the absence of cross-document repetition, single document summarizers for news have not been able to consistently outperform the strong beginning-of-article baseline.

Despite sequences being core to NLP, scant work has considered how to handle noisy sequence labels from multiple annotators for the same text. Given such annotations, we consider two complementary tasks: (1) aggregating sequential crowd labels to infer a best single set of consensus annotations; and (2) using crowd annotations as training data for a model that can predict sequences in unannotated text. For aggregation, we propose a novel Hidden Markov Model variant. To predict sequences in unannotated text, we propose a neural approach using Long Short Term Memory. We evaluate a suite of methods across two different applications and text genres: Named-Entity Recognition in news articles and Information Extraction from biomedical abstracts. Results show improvement over strong baselines. Our source code and data are available online.

We introduce improved guidelines for annotation of sentence specificity, addressing the issues encountered in prior work. Our annotation provides judgements of sentences in context. Rather than binary judgements, we introduce a specificity scale which accommodates nuanced judgements. Our augmented annotation procedure also allows us to define where in the discourse context the lack of specificity can be resolved. In addition, the cause of the underspecification is annotated in the form of free text questions. We present results from a pilot annotation with this new scheme and demonstrate good inter-annotator agreement. We found that the lack of specificity distributes evenly among immediate prior context, long distance prior context and no prior context. We find that missing details that are not resolved in the the prior context are more likely to trigger questions about the reason behind events, “why” and “how”. Our data is accessible at http://www.cis.upenn.edu/~nlp/corpora/lrec16spec.html

In the period since 2004, many novel sophisticated approaches for generic multi-document summarization have been developed. Intuitive simple approaches have also been shown to perform unexpectedly well for the task. Yet it is practically impossible to compare the existing approaches directly, because systems have been evaluated on different datasets, with different evaluation measures, against different sets of comparison systems. Here we present a corpus of summaries produced by several state-of-the-art extractive summarization systems or by popular baseline systems. The inputs come from the 2004 DUC evaluation, the latest year in which generic summarization was addressed in a shared task. We use the same settings for ROUGE automatic evaluation to compare the systems directly and analyze the statistical significance of the differences in performance. We show that in terms of average scores the state-of-the-art systems appear similar but that in fact they produce very different summaries. Our corpus will facilitate future research on generic summarization and motivates the need for development of more sensitive evaluation measures and for approaches to system combination in summarization.

Great writing is rare and highly admired. Readers seek out articles that are beautifully written, informative and entertaining. Yet information-access technologies lack capabilities for predicting article quality at this level. In this paper we present first experiments on article quality prediction in the science journalism domain. We introduce a corpus of great pieces of science journalism, along with typical articles from the genre. We implement features to capture aspects of great writing, including surprising, visual and emotional content, as well as general features related to discourse organization and sentence structure. We show that the distinction between great and typical articles can be detected fairly accurately, and that the entire spectrum of our features contribute to the distinction.

We present a corpus of sentences from news articles that are annotated as general or specific. We employed annotators on Amazon Mechanical Turk to mark sentences from three kinds of news articles―reports on events, finance news and science journalism. We introduce the resulting corpus, with focus on annotator agreement, proportion of general/specific sentences in the articles and results for automatic classification of the two sentence types.