We introduce Holmes , a new benchmark designed to assess language models’ (LMs’) linguistic competence —their unconscious understanding of linguistic phenomena. Specifically, we use classifier-based probing to examine LMs’ internal representations regarding distinct linguistic phenomena (e.g., part-of-speech tagging). As a result, we meet recent calls to disentangle LMs’ linguistic competence from other cognitive abilities, such as following instructions in prompting-based evaluations. Composing Holmes , we review over 270 probing studies and include more than 200 datasets to assess syntax, morphology, semantics, reasoning, and discourse phenomena. Analyzing over 50 LMs reveals that, aligned with known trends, their linguistic competence correlates with model size. However, surprisingly, model architecture and instruction tuning also significantly influence performance, particularly in morphology and syntax . Finally, we propose FlashHolmes , a streamlined version that reduces the computation load while maintaining high-ranking precision.

Automated fact-checking systems verify claims against evidence to predict their veracity. In real-world scenarios, the retrieved evidence may not unambiguously support or refute the claim and yield conflicting but valid interpretations. Existing fact-checking datasets assume that the models developed with them predict a single veracity label for each claim, thus discouraging the handling of such ambiguity. To address this issue we present AmbiFC , 1 a fact-checking dataset with 10k claims derived from real-world information needs. It contains fine-grained evidence annotations of 50k passages from 5k Wikipedia pages. We analyze the disagreements arising from ambiguity when comparing claims against evidence in AmbiFC , observing a strong correlation of annotator disagreement with linguistic phenomena such as underspecification and probabilistic reasoning. We develop models for predicting veracity handling this ambiguity via soft labels, and find that a pipeline that learns the label distribution for sentence-level evidence selection and veracity prediction yields the best performance. We compare models trained on different subsets of AmbiFC and show that models trained on the ambiguous instances perform better when faced with the identified linguistic phenomena.

Recent work in natural language processing (NLP) has yielded appealing results from scaling model parameters and training data; however, using only scale to improve performance means that resource consumption also grows. Such resources include data, time, storage, or energy, all of which are naturally limited and unevenly distributed. This motivates research into efficient methods that require fewer resources to achieve similar results. This survey synthesizes and relates current methods and findings in efficient NLP. We aim to provide both guidance for conducting NLP under limited resources, and point towards promising research directions for developing more efficient methods.

Despite extensive research efforts in recent years, computational argumentation (CA) remains one of the most challenging areas of natural language processing. The reason for this is the inherent complexity of the cognitive processes behind human argumentation, which integrate a plethora of different types of knowledge, ranging from topic-specific facts and common sense to rhetorical knowledge. The integration of knowledge from such a wide range in CA requires modeling capabilities far beyond many other natural language understanding tasks. Existing research on mining, assessing, reasoning over, and generating arguments largely acknowledges that much more knowledge is needed to accurately model argumentation computationally. However, a systematic overview of the types of knowledge introduced in existing CA models is missing, hindering targeted progress in the field. Adopting the operational definition of knowledge as any task-relevant normative information not provided as input, the survey paper at hand fills this gap by (1) proposing a taxonomy of types of knowledge required in CA tasks, (2) systematizing the large body of CA work according to the reliance on and exploitation of these knowledge types for the four main research areas in CA, and (3) outlining and discussing directions for future research efforts in CA.

Current state-of-the-art approaches to cross- modal retrieval process text and visual input jointly, relying on Transformer-based architectures with cross-attention mechanisms that attend over all words and objects in an image. While offering unmatched retrieval performance, such models: 1) are typically pretrained from scratch and thus less scalable, 2) suffer from huge retrieval latency and inefficiency issues, which makes them impractical in realistic applications. To address these crucial gaps towards both improved and efficient cross- modal retrieval, we propose a novel fine-tuning framework that turns any pretrained text-image multi-modal model into an efficient retrieval model. The framework is based on a cooperative retrieve-and-rerank approach that combines: 1) twin networks (i.e., a bi-encoder) to separately encode all items of a corpus, enabling efficient initial retrieval, and 2) a cross-encoder component for a more nuanced (i.e., smarter) ranking of the retrieved small set of items. We also propose to jointly fine- tune the two components with shared weights, yielding a more parameter-efficient model. Our experiments on a series of standard cross-modal retrieval benchmarks in monolingual, multilingual, and zero-shot setups, demonstrate improved accuracy and huge efficiency benefits over the state-of-the-art cross- encoders. 1

For many NLP applications, such as question answering and summarization, the goal is to select the best solution from a large space of candidates to meet a particular user’s needs. To address the lack of user or task-specific training data, we propose an interactive text ranking approach that actively selects pairs of candidates, from which the user selects the best. Unlike previous strategies, which attempt to learn a ranking across the whole candidate space, our method uses Bayesian optimization to focus the user’s labeling effort on high quality candidates and integrate prior knowledge to cope better with small data scenarios. We apply our method to community question answering (cQA) and extractive multidocument summarization, finding that it significantly outperforms existing interactive approaches. We also show that the ranking function learned by our method is an effective reward function for reinforcement learning, which improves the state of the art for interactive summarization.

Recent graph-to-text models generate text from graph-based data using either global or local aggregation to learn node representations. Global node encoding allows explicit communication between two distant nodes, thereby neglecting graph topology as all nodes are directly connected. In contrast, local node encoding considers the relations between neighbor nodes capturing the graph structure, but it can fail to capture long-range relations. In this work, we gather both encoding strategies, proposing novel neural models that encode an input graph combining both global and local node contexts, in order to learn better contextualized node embeddings. In our experiments, we demonstrate that our approaches lead to significant improvements on two graph-to-text datasets achieving BLEU scores of 18.01 on the AGENDA dataset, and 63.69 on the WebNLG dataset for seen categories, outperforming state-of-the-art models by 3.7 and 3.1 points, respectively. 1

We introduce a scalable Bayesian preference learning method for identifying convincing arguments in the absence of gold-standard ratings or rankings. In contrast to previous work, we avoid the need for separate methods to perform quality control on training data, predict rankings and perform pairwise classification. Bayesian approaches are an effective solution when faced with sparse or noisy training data, but have not previously been used to identify convincing arguments. One issue is scalability, which we address by developing a stochastic variational inference method for Gaussian process (GP) preference learning. We show how our method can be applied to predict argument convincingness from crowdsourced data, outperforming the previous state-of-the-art, particularly when trained with small amounts of unreliable data. We demonstrate how the Bayesian approach enables more effective active learning, thereby reducing the amount of data required to identify convincing arguments for new users and domains. While word embeddings are principally used with neural networks, our results show that word embeddings in combination with linguistic features also benefit GPs when predicting argument convincingness.

Extracting the information from text when an event happened is challenging. Documents do not only report on current events, but also on past events as well as on future events. Often, the relevant time information for an event is scattered across the document. In this paper we present a novel method to automatically anchor events in time. To our knowledge it is the first approach that takes temporal information from the complete document into account. We created a decision tree that applies neural network based classifiers at its nodes. We use this tree to incrementally infer, in a stepwise manner, at which time frame an event happened. We evaluate the approach on the TimeBank-EventTime Corpus (Reimers et al., 2016) achieving an accuracy of 42.0% compared to an inter-annotator agreement (IAA) of 56.7%. For events that span over a single day we observe an accuracy improvement of 33.1 points compared to the state-of-the-art CAEVO system (Chambers et al., 2014). Without retraining, we apply this model to the SemEval-2015 Task 4 on automatic timeline generation and achieve an improvement of 4.01 points F 1 -score compared to the state-of-the-art. Our code is publically available.

We present a new approach for generating role-labeled training data using Linked Lexical Resources, i.e., integrated lexical resources that combine several resources (e.g., Word-Net, FrameNet, Wiktionary) by linking them on the sense or on the role level. Unlike resource-based supervision in relation extraction, we focus on complex linguistic annotations, more specifically FrameNet senses and roles. The automatically labeled training data ( www.ukp.tu-darmstadt.de/knowledge-based-srl/ ) are evaluated on four corpora from different domains for the tasks of word sense disambiguation and semantic role classification. Results show that classifiers trained on our generated data equal those resulting from a standard supervised setting.

Language proficiency tests are used to evaluate and compare the progress of language learners. We present an approach for automatic difficulty prediction of C-tests that performs on par with human experts. On the basis of detailed analysis of newly collected data, we develop a model for C-test difficulty introducing four dimensions: solution difficulty, candidate ambiguity, inter-gap dependency, and paragraph difficulty. We show that cues from all four dimensions contribute to C-test difficulty.

In this paper, we present Dijkstra-WSA, a novel graph-based algorithm for word sense alignment. We evaluate it on four different pairs of lexical-semantic resources with different characteristics (WordNet-OmegaWiki, WordNet-Wiktionary, GermaNet-Wiktionary and WordNet-Wikipedia) and show that it achieves competitive performance on 3 out of 4 datasets. Dijkstra-WSA outperforms the state of the art on every dataset if it is combined with a back-off based on gloss similarity. We also demonstrate that Dijkstra-WSA is not only flexibly applicable to different resources but also highly parameterizable to optimize for precision or recall.