Frames of Communication (FoCs) are ubiquitous in social media discourse. They define what counts as a problem, diagnose what is causing the problem, elicit moral judgments and imply remedies for resolving the problem. Most research on automatic frame detection involved the recognition of the problems addressed by frames, but did not consider the articulation of frames. Articulating an FoC involves reasoning with salient problems, their cause and eventual solution. In this paper we present a method for Discovering and Articulating FoCs (DA-FoC) that relies on a combination of Chain-of-Thought prompting of large language models (LLMs) with In-Context Active Curriculum Learning. Very promising evaluation results indicate that 86.72% of the FoCs encoded by communication experts on the same reference dataset were also uncovered by DA-FoC. Moreover, DA-FoC uncovered many new FoCs, which escaped the experts. Interestingly, 55.1% of the known FoCs were judged as being better articulated than the human-written ones, while 93.8% of the new FoCs were judged as having sound rationale and being clearly articulated.

Stance detection enables the inference of attitudes from human communications. Automatic stance identification was mostly cast as a classification problem. However, stance decisions involve complex judgments, which can be nowadays generated by prompting Large Language Models (LLMs). In this paper we present a new method for stance identification which (1) relies on a new prompting framework, called Tree-of-Counterfactual prompting; (2) operates not only on textual communications, but also on images; (3) allows more than one stance object type; and (4) requires no examples of stance attribution, thus it is a “Tabula Rasa” Zero-Shot Stance Detection (TR-ZSSD) method. Our experiments indicate surprisingly promising results, outperforming fine-tuned stance detection systems.

Stance as an expression of an author’s standpoint and as a means of communication has long been studied by computational linguists. Automatically identifying the stance of a subject toward an object is an active area of research in natural language processing. Significant work has employed topics and claims as the object of stance, with frames of communication becoming more recently considered as alternative objects of stance. However, little attention has been paid to finding what are the benefits and what are the drawbacks when inferring the stance of a text towards different possible stance objects. In this paper we seek to answer this question by analyzing the implied knowledge and the judgments required when deciding the stance of a text towards each stance object type. Our analysis informed experiments with models capable of inferring the stance of a text towards any of the stance object types considered, namely topics, claims, and frames of communication. Experiments clearly indicate that it is best to infer the stance of a text towards a frame of communication, rather than a claim or a topic. It is also better to infer the stance of a text towards a claim rather than a topic. Therefore we advocate that rather than continuing efforts to annotate the stance of texts towards topics, it is better to use those efforts to produce annotations towards frames of communication. These efforts will allow us to better capture the stance towards claims and topics as well.

Frames of communication are often evoked in multimedia documents. When an author decides to add an image to a text, one or both of the modalities may evoke a communication frame. Moreover, when evoking the frame, the author also conveys her/his stance towards the frame. Until now, determining if the author is in favor of, against or has no stance towards the frame was performed automatically only when processing texts. This is due to the absence of stance annotations on multimedia documents. In this paper we introduce MMVax-Stance, a dataset of 11,300 multimedia documents retrieved from social media, which have stance annotations towards 113 different frames of communication. This dataset allowed us to experiment with several models of multimedia stance detection, which revealed important interactions between texts and images in the inference of stance towards communication frames. When inferring the text/image relations, a set of 46,606 synthetic examples of multimodal documents with known stance was generated. This greatly impacted the quality of identifying multimedia stance, yielding an improvement of 20% in F1-score.

Billions of COVID-19 vaccines have been administered, but many remain hesitant. Misinformation about the COVID-19 vaccines and other vaccines, propagating on social media, is believed to drive hesitancy towards vaccination. The ability to automatically recognize misinformation targeting vaccines on Twitter depends on the availability of data resources. In this paper we present VaccineLies, a large collection of tweets propagating misinformation about two vaccines: the COVID-19 vaccines and the Human Papillomavirus (HPV) vaccines. Misinformation targets are organized in vaccine-specific taxonomies, which reveal the misinformation themes and concerns. The ontological commitments of the misinformation taxonomies provide an understanding of which misinformation themes and concerns dominate the discourse about the two vaccines covered in VaccineLies. The organization into training, testing and development sets of VaccineLies invites the development of novel supervised methods for detecting misinformation on Twitter and identifying the stance towards it. Furthermore, VaccineLies can be a stepping stone for the development of datasets focusing on misinformation targeting additional vaccines.

Extracting structured knowledge involving self-reported events related to the COVID-19 pandemic from Twitter has the potential to inform surveillance systems that play a critical role in public health. The event extraction challenge presented by the W-NUT 2020 Shared Task 3 focused on the identification of five types of events relevant to the COVID-19 pandemic and their respective set of pre-defined slots encoding demographic, epidemiological, clinical as well as spatial, temporal or subjective knowledge. Our participation in the challenge led to the design of a neural architecture for jointly identifying all Event Slots expressed in a tweet relevant to an event of interest. This architecture uses COVID-Twitter-BERT as the pre-trained language model. In addition, to learn text span embeddings for each Event Slot, we relied on a special case of Hopfield Networks, namely Hopfield pooling. The results of the shared task evaluation indicate that our system performs best when it is trained on a larger dataset, while it remains competitive when training on smaller datasets.

Brain signals are captured by clinical electroencephalography (EEG) which is an excellent tool for probing neural function. When EEG tests are performed, a textual EEG report is generated by the neurologist to document the findings, thus using language that describes the brain signals and its clinical correlations. Even with the impetus provided by the BRAIN initiative (brainitititive.nih.gov), there are no annotations available in texts that capture language describing the brain activities and their correlations with various pathologies. In this paper we describe an annotation effort carried out on a large corpus of EEG reports, providing examples of EEG-specific and clinically relevant concepts. In addition, we detail our annotation schema for brain signal attributes. We also discuss the resulting annotation of long-distance relations between concepts in EEG reports. By exemplifying a self-attention joint-learning to predict similar annotations in the EEG report corpus, we discuss the promising results, hoping that our effort will inform the design of novel knowledge capture techniques that will include the language of brain signals.

Our ability to understand language often relies on common-sense knowledge ― background information the speaker can assume is known by the reader. Similarly, our comprehension of the language used in complex domains relies on access to domain-specific knowledge. Capturing common-sense and domain-specific knowledge can be achieved by taking advantage of recent advances in open information extraction (IE) techniques and, more importantly, of knowledge embeddings, which are multi-dimensional representations of concepts and relations. Building a knowledge graph for representing common-sense knowledge in which concepts discerned from noun phrases are cast as vertices and lexicalized relations are cast as edges leads to learning the embeddings of common-sense knowledge accounting for semantic compositionality as well as implied knowledge. Common-sense knowledge is acquired from a vast collection of blogs and books as well as from WordNet. Similarly, medical knowledge is learned from two large sets of electronic health records. The evaluation results of these two forms of knowledge are promising: the same knowledge acquisition methodology based on learning knowledge embeddings works well both for common-sense knowledge and for medical knowledge Interestingly, the common-sense knowledge that we have acquired was evaluated as being less neutral than than the medical knowledge, as it often reflected the opinion of the knowledge utterer. In addition, the acquired medical knowledge was evaluated as more plausible than the common-sense knowledge, reflecting the complexity of acquiring common-sense knowledge due to the pragmatics and economicity of language.

Electronic Medical Records (EMRs) encode an extraordinary amount of medical knowledge. Collecting and interpreting this knowledge, however, belies a significant level of clinical understanding. Automatically capturing the clinical information is crucial for performing comparative effectiveness research. In this paper, we present a data-driven approach to model semantic dependencies between medical concepts, qualified by the beliefs of physicians. The dependencies, captured in a patient cohort graph of clinical pictures and therapies is further refined into a probabilistic graphical model which enables efficient inference of patient-centered treatment or test recommendations (based on probabilities). To perform inference on the graphical model, we describe a technique of smoothing the conditional likelihood of medical concepts by their semantically-similar belief values. The experimental results, as compared against clinical guidelines are very promising.

The rise of micro-blogging in recent years has resulted in significant access to emotion-laden text. Unlike emotion expressed in other textual sources (e.g., blogs, quotes in newswire, email, product reviews, or even clinical text), micro-blogs differ by (1) placing a strict limit on length, resulting radically in new forms of emotional expression, and (2) encouraging users to express their daily thoughts in real-time, often resulting in far more emotion statements than might normally occur. In this paper, we introduce a corpus collected from Twitter with annotated micro-blog posts (or “tweets”) annotated at the tweet-level with seven emotions: ANGER, DISGUST, FEAR, JOY, LOVE, SADNESS, and SURPRISE. We analyze how emotions are distributed in the data we annotated and compare it to the distributions in other emotion-annotated corpora. We also used the annotated corpus to train a classifier that automatically discovers the emotions in tweets. In addition, we present an analysis of the linguistic style used for expressing emotions our corpus. We hope that these observations will lead to the design of novel emotion detection techniques that account for linguistic style and psycholinguistic theories.

A significant amount of spatial information in textual documents is hidden within the relationship between events. While humans have an intuitive understanding of these relationships that allow us to recover an object's or event's location, currently no annotated data exists to allow automatic discovery of spatial containment relations between events. We present our process for building such a corpus of manually annotated spatial relations between events. Events form complex predicate-argument structures that model the participants in the event, their roles, as well as the temporal and spatial grounding. In addition, events are not presented in isolation in text; there are explicit and implicit interactions between events that often participate in event structures. In this paper, we focus on five spatial containment relations that may exist between events: (1) SAME, (2) CONTAINS, (3) OVERLAPS, (4) NEAR, and (5) DIFFERENT. Using the transitive closure across these spatial relations, the implicit location of many events and their participants can be discovered. We discuss our annotation schema for spatial containment relations, placing it within the pre-existing theories of spatial representation. We also discuss our annotation guidelines for maintaining annotation quality as well as our process for augmenting SpatialML with spatial containment relations between events. Additionally, we outline some baseline experiments to evaluate the feasibility of developing supervised systems based on this corpus. These results indicate that although the task is challenging, automated methods are capable of discovering spatial containment relations between events.

This paper presents a corpus of annotated motion events and their event structure. We consider motion events triggered by a set of motion evoking words and contemplate both literal and figurative interpretations of them. Figurative motion events are extracted into the same event structure but are marked as figurative in the corpus. To represent the event structure of motion, we use the FrameNet annotation standard, which encodes motion in over 70 frames. In order to acquire a diverse set of texts that are different from FrameNet's, we crawled blog and news feeds for five different domains: sports, newswire, finance, military, and gossip. We then annotated these documents with an automatic FrameNet parser. Its output was manually corrected to account for missing and incorrect frames as well as missing and incorrect frame elements. The corpus, UTD-MotionEvent, may act as a resource for semantic parsing, detection of figurative language, spatial reasoning, and other tasks.

In this paper, we present a linguistic resource that annotates event structures in texts. We consider an event structure as a collection of events that interact with each other in a given situation. We interpret the interactions between events as event relations. In this regard, we propose and annotate a set of six relations that best capture the concept of event structure. These relations are: subevent, reason, purpose, enablement, precedence and related. A document from this resource can encode multiple event structures and an event structure can be described across multiple documents. In order to unify event structures, we also annotate inter- and intra-document event coreference. Moreover, we provide methodologies for automatic discovery of event structures from texts. First, we group the events that constitute an event structure into event clusters and then, we use supervised learning frameworks to classify the relations that exist between events from the same cluster

Generating answers to complex questions in the form of multi-document summaries requires access to question decomposition methods. In this paper we present three methods for decomposing complex questions and we evaluate their impact on the responsiveness of the answers they enable.

Answering questions that ask about temporal information involves several forms of inference. In order to develop question answering capabilities that benefit from temporal inference, we believe that a large corpus of questions and answers that are discovered based on temporal information should be available. This paper describes our methodology for creating AnswerTime-Bank, a large corpus of questions and answers on which Question Answering systems can operate using complex temporal inference.

This paper describes a novel clustering-based text summarization system that uses Multiple Sequence Alignment to improve the alignment of sentences within topic clusters. While most current clustering-based summarization systems base their summaries only on the common information contained in a collection of highly-related sentences, our system constructs more informative summaries that incorporate both the redundant and unique contributions of the sentences in the cluster. When evaluated using ROUGE, the summaries produced by our system represent a substantial improvement over the baseline, which is at 63% of the human performance.