Fred Morstatter

Omar Shaikh, Caleb Ziems, William Held et al. · gatech

Pragmatic reference enables efficient interpersonal communication. Prior work uses simple reference games to test models of pragmatic reasoning, often with unidentified speakers and listeners. In practice, however, speakers' sociocultural background shapes their pragmatic assumptions. For example, readers of this paper assume NLP refers to "Natural Language Processing," and not "Neuro-linguistic Programming." This work introduces the Cultural Codes dataset, which operationalizes sociocultural pragmatic inference in a simple word reference game. Cultural Codes is based on the multi-turn collaborative two-player game, Codenames Duet. Our dataset consists of 794 games with 7,703 turns, distributed across 153 unique players. Alongside gameplay, we collect information about players' personalities, values, and demographics. Utilizing theories of communication and pragmatics, we predict each player's actions via joint modeling of their sociocultural priors and the game context. Our experiments show that accounting for background characteristics significantly improves model performance for tasks related to both clue giving and guessing, indicating that sociocultural priors play a vital role in gameplay decisions.

Ninareh Mehrabi, Ahmad Beirami, Fred Morstatter et al.

Warning: this paper contains content that maybe offensive or upsetting. Recent research in Natural Language Processing (NLP) has advanced the development of various toxicity detection models with the intention of identifying and mitigating toxic language from existing systems. Despite the abundance of research in this area, less attention has been given to adversarial attacks that force the system to generate toxic language and the defense against them. Existing work to generate such attacks is either based on human-generated attacks which is costly and not scalable or, in case of automatic attacks, the attack vector does not conform to human-like language, which can be detected using a language model loss. In this work, we propose attacks against conversational agents that are imperceptible, i.e., they fit the conversation in terms of coherency, relevancy, and fluency, while they are effective and scalable, i.e., they can automatically trigger the system into generating toxic language. We then propose a defense mechanism against such attacks which not only mitigates the attack but also attempts to maintain the conversational flow. Through automatic and human evaluations, we show that our defense is effective at avoiding toxic language generation even against imperceptible toxicity triggers while the generated language fits the conversation in terms of coherency and relevancy. Lastly, we establish the generalizability of such a defense mechanism on language generation models beyond conversational agents.

Abel Salinas, Parth Vipul Shah, Yuzhong Huang et al.

Large Language Models (LLMs) have seen widespread deployment in various real-world applications. Understanding these biases is crucial to comprehend the potential downstream consequences when using LLMs to make decisions, particularly for historically disadvantaged groups. In this work, we propose a simple method for analyzing and comparing demographic bias in LLMs, through the lens of job recommendations. We demonstrate the effectiveness of our method by measuring intersectional biases within ChatGPT and LLaMA, two cutting-edge LLMs. Our experiments primarily focus on uncovering gender identity and nationality bias; however, our method can be extended to examine biases associated with any intersection of demographic identities. We identify distinct biases in both models toward various demographic identities, such as both models consistently suggesting low-paying jobs for Mexican workers or preferring to recommend secretarial roles to women. Our study highlights the importance of measuring the bias of LLMs in downstream applications to understand the potential for harm and inequitable outcomes.

Negar Mokhberian, Myrl G. Marmarelis, Frederic R. Hopp et al.

Supervised classification heavily depends on datasets annotated by humans. However, in subjective tasks such as toxicity classification, these annotations often exhibit low agreement among raters. Annotations have commonly been aggregated by employing methods like majority voting to determine a single ground truth label. In subjective tasks, aggregating labels will result in biased labeling and, consequently, biased models that can overlook minority opinions. Previous studies have shed light on the pitfalls of label aggregation and have introduced a handful of practical approaches to tackle this issue. Recently proposed multi-annotator models, which predict labels individually per annotator, are vulnerable to under-determination for annotators with few samples. This problem is exacerbated in crowdsourced datasets. In this work, we propose \textbf{Annotator Aware Representations for Texts (AART)} for subjective classification tasks. Our approach involves learning representations of annotators, allowing for exploration of annotation behaviors. We show the improvement of our method on metrics that assess the performance on capturing individual annotators' perspectives. Additionally, we demonstrate fairness metrics to evaluate our model's equability of performance for marginalized annotators compared to others.

Myrl G. Marmarelis, Greg Ver Steeg, Aram Galstyan et al.

Causal inference of exact individual treatment outcomes in the presence of hidden confounders is rarely possible. Recent work has extended prediction intervals with finite-sample guarantees to partially identifiable causal outcomes, by means of a sensitivity model for hidden confounding. In deep learning, predictors can exploit their inductive biases for better generalization out of sample. We argue that the structure inherent to a deep ensemble should inform a tighter partial identification of the causal outcomes that they predict. We therefore introduce an approach termed Caus-Modens, for characterizing causal outcome intervals by modulated ensembles. We present a simple approach to partial identification using existing causal sensitivity models and show empirically that Caus-Modens gives tighter outcome intervals, as measured by the necessary interval size to achieve sufficient coverage. The last of our three diverse benchmarks is a novel usage of GPT-4 for observational experiments with unknown but probeable ground truth.

Negar Mokhberian, Frederic R. Hopp, Bahareh Harandizadeh et al.

Morality plays an important role in culture, identity, and emotion. Recent advances in natural language processing have shown that it is possible to classify moral values expressed in text at scale. Morality classification relies on human annotators to label the moral expressions in text, which provides training data to achieve state-of-the-art performance. However, these annotations are inherently subjective and some of the instances are hard to classify, resulting in noisy annotations due to error or lack of agreement. The presence of noise in training data harms the classifier's ability to accurately recognize moral foundations from text. We propose two metrics to audit the noise of annotations. The first metric is entropy of instance labels, which is a proxy measure of annotator disagreement about how the instance should be labeled. The second metric is the silhouette coefficient of a label assigned by an annotator to an instance. This metric leverages the idea that instances with the same label should have similar latent representations, and deviations from collective judgments are indicative of errors. Our experiments on three widely used moral foundations datasets show that removing noisy annotations based on the proposed metrics improves classification performance.

Berk Atil, Rebecca J. Passonneau, Fred Morstatter

Large language models (LLMs) undergo safety alignment after training and tuning, yet recent work shows that safety can be bypassed through jailbreak attacks. While many jailbreaks and defenses exist, their cross-lingual generalization remains underexplored. This paper presents the first systematic multilingual evaluation of jailbreaks and defenses across ten languages -- spanning high-, medium-, and low-resource languages -- using six LLMs on HarmBench and AdvBench. We assess two jailbreak types: logical-expression-based and adversarial-prompt-based. For both types, attack success and defense robustness vary across languages: high-resource languages are safer under standard queries but more vulnerable to adversarial ones. Simple defenses can be effective, but are language- and model-dependent. These findings call for language-aware and cross-lingual safety benchmarks for LLMs.

Harsh Sakhrani, Naseela Pervez, Anirudh Ravi Kumar et al.

It is desirable to coarsely classify short scientific texts, such as grant or publication abstracts, for strategic insight or research portfolio management. These texts efficiently transmit dense information to experts possessing a rich body of knowledge to aid interpretation. Yet this task is remarkably difficult to automate because of brevity and the absence of context. To address this gap, we have developed a novel approach to generate and appropriately assign coarse domain-specific labels. We show that a Large Language Model (LLM) can provide metadata essential to the task, in a process akin to the augmentation of supplemental knowledge representing human intuition, and propose a workflow. As a pilot study, we use a corpus of award abstracts from the National Aeronautics and Space Administration (NASA). We develop new assessment tools in concert with established performance metrics.

Abel Salinas, Louis Penafiel, Robert McCormack et al.

Large language models (LLMs) have garnered significant attention for their remarkable performance in a continuously expanding set of natural language processing tasks. However, these models have been shown to harbor inherent societal biases, or stereotypes, which can adversely affect their performance in their many downstream applications. In this paper, we introduce a novel, purely prompt-based approach to uncover hidden stereotypes within any arbitrary LLM. Our approach dynamically generates a knowledge representation of internal stereotypes, enabling the identification of biases encoded within the LLM's internal knowledge. By illuminating the biases present in LLMs and offering a systematic methodology for their analysis, our work contributes to advancing transparency and promoting fairness in natural language processing systems.

Kian Ahrabian, Zhivar Sourati, Kexuan Sun et al.

While large language models (LLMs) are still being adopted to new domains and utilized in novel applications, we are experiencing an influx of the new generation of foundation models, namely multi-modal large language models (MLLMs). These models integrate verbal and visual information, opening new possibilities to demonstrate more complex reasoning abilities at the intersection of the two modalities. However, despite the revolutionizing prospect of MLLMs, our understanding of their reasoning abilities is limited. In this study, we assess the nonverbal abstract reasoning abilities of open-source and closed-source MLLMs using variations of Raven's Progressive Matrices. Our experiments reveal the challenging nature of such problems for MLLMs while showcasing the immense gap between open-source and closed-source models. We also uncover critical shortcomings of visual and textual perceptions, subjecting the models to low-performance ceilings. Finally, to improve MLLMs' performance, we experiment with different methods, such as Chain-of-Thought prompting, leading to a significant (up to 100%) boost in performance. Our code and datasets are available at https://github.com/usc-isi-i2/isi-mmlm-rpm.

Yuzhong Huang, Chen Liu, Ji Hou et al.

We present UniPlane, a novel method that unifies plane detection and reconstruction from posed monocular videos. Unlike existing methods that detect planes from local observations and associate them across the video for the final reconstruction, UniPlane unifies both the detection and the reconstruction tasks in a single network, which allows us to directly optimize final reconstruction quality and fully leverage temporal information. Specifically, we build a Transformers-based deep neural network that jointly constructs a 3D feature volume for the environment and estimates a set of per-plane embeddings as queries. UniPlane directly reconstructs the 3D planes by taking dot products between voxel embeddings and the plane embeddings followed by binary thresholding. Extensive experiments on real-world datasets demonstrate that UniPlane outperforms state-of-the-art methods in both plane detection and reconstruction tasks, achieving +4.6 in F-score in geometry as well as consistent improvements in other geometry and segmentation metrics.

Jundong Li, Kewei Cheng, Suhang Wang et al.

Feature selection, as a data preprocessing strategy, has been proven to be effective and efficient in preparing data (especially high-dimensional data) for various data mining and machine learning problems. The objectives of feature selection include: building simpler and more comprehensible models, improving data mining performance, and preparing clean, understandable data. The recent proliferation of big data has presented some substantial challenges and opportunities to feature selection. In this survey, we provide a comprehensive and structured overview of recent advances in feature selection research. Motivated by current challenges and opportunities in the era of big data, we revisit feature selection research from a data perspective and review representative feature selection algorithms for conventional data, structured data, heterogeneous data and streaming data. Methodologically, to emphasize the differences and similarities of most existing feature selection algorithms for conventional data, we categorize them into four main groups: similarity based, information theoretical based, sparse learning based and statistical based methods. To facilitate and promote the research in this community, we also present an open-source feature selection repository that consists of most of the popular feature selection algorithms (\url{http://featureselection.asu.edu/}). Also, we use it as an example to show how to evaluate feature selection algorithms. At the end of the survey, we present a discussion about some open problems and challenges that require more attention in future research.

Abel Salinas, Fred Morstatter

Large Language Models (LLMs) are regularly being used to label data across many domains and for myriad tasks. By simply asking the LLM for an answer, or ``prompting,'' practitioners are able to use LLMs to quickly get a response for an arbitrary task. This prompting is done through a series of decisions by the practitioner, from simple wording of the prompt, to requesting the output in a certain data format, to jailbreaking in the case of prompts that address more sensitive topics. In this work, we ask: do variations in the way a prompt is constructed change the ultimate decision of the LLM? We answer this using a series of prompt variations across a variety of text classification tasks. We find that even the smallest of perturbations, such as adding a space at the end of a prompt, can cause the LLM to change its answer. Further, we find that requesting responses in XML and commonly used jailbreaks can have cataclysmic effects on the data labeled by LLMs.

Rebecca Dorn, Christina Chance, Casandra Rusti et al.

Automated emotion detection is widely used in applications ranging from well-being monitoring to high-stakes domains like mental health and hiring. However, models often rely on annotations that reflect dominant cultural norms, limiting model ability to recognize emotional expression in dialects often excluded from training data distributions, such as African American Vernacular English (AAVE). This study examines emotion recognition model performance on AAVE compared to General American English (GAE). We analyze 2.7 million tweets geo-tagged within Los Angeles. Texts are scored for strength of AAVE using computational approximations of dialect features. Annotations of emotion presence and intensity are collected on a dataset of 875 tweets with both high and low AAVE densities. To assess model accuracy on a task as subjective as emotion perception, we calculate community-informed "silver" labels where AAVE-dense tweets are labeled by African American, AAVE-fluent (ingroup) annotators. On our labeled sample, GPT and BERT-based models exhibit false positive prediction rates of anger on AAVE more than double than on GAE. SpanEmo, a popular text-based emotion model, increases false positive rates of anger from 25 percent on GAE to 60 percent on AAVE. Additionally, a series of linear regressions reveals that models and non-ingroup annotations are significantly more correlated with profanity-based AAVE features than ingroup annotations. Linking Census tract demographics, we observe that neighborhoods with higher proportions of African American residents are associated with higher predictions of anger (Pearson's correlation r = 0.27) and lower joy (r = -0.10). These results find an emergent safety issue of emotion AI reinforcing racial stereotypes through biased emotion classification. We emphasize the need for culturally and dialect-informed affective computing systems.

James Y. Huang, Wenxuan Zhou, Fei Wang et al. · microsoft-research

Despite the strong capabilities of Large Language Models (LLMs) to acquire knowledge from their training corpora, the memorization of sensitive information in the corpora such as copyrighted, biased, and private content has led to ethical and legal concerns. In response to these challenges, unlearning has emerged as a potential remedy for LLMs affected by problematic training data. However, previous unlearning techniques are either not applicable to black-box LLMs due to required access to model internal weights, or violate data protection principles by retaining sensitive data for inference-time correction. We propose δ-Unlearning, an offset unlearning framework for black-box LLMs. Instead of tuning the black-box LLM itself, δ-Unlearning learns the logit offset needed for unlearning by contrasting the logits from a pair of smaller models. Experiments demonstrate that δ- Unlearning can effectively unlearn target data while maintaining similar or even stronger performance on general out-of-forget-scope tasks. δ-Unlearning also effectively incorporates different unlearning algorithms, making our approach a versatile solution to adapting various existing unlearning algorithms to black-box LLMs.

Daniel M. Benjamin, Fred Morstatter, Ali E. Abbas et al. · stanford

Sound decision-making relies on accurate prediction for tangible outcomes ranging from military conflict to disease outbreaks. To improve crowdsourced forecasting accuracy, we developed SAGE, a hybrid forecasting system that combines human and machine generated forecasts. The system provides a platform where users can interact with machine models and thus anchor their judgments on an objective benchmark. The system also aggregates human and machine forecasts weighting both for propinquity and based on assessed skill while adjusting for overconfidence. We present results from the Hybrid Forecasting Competition (HFC) - larger than comparable forecasting tournaments - including 1085 users forecasting 398 real-world forecasting problems over eight months. Our main result is that the hybrid system generated more accurate forecasts compared to a human-only baseline which had no machine generated predictions. We found that skilled forecasters who had access to machine-generated forecasts outperformed those who only viewed historical data. We also demonstrated the inclusion of machine-generated forecasts in our aggregation algorithms improved performance, both in terms of accuracy and scalability. This suggests that hybrid forecasting systems, which potentially require fewer human resources, can be a viable approach for maintaining a competitive level of accuracy over a larger number of forecasting questions.

Rebecca Dorn, Lee Kezar, Fred Morstatter et al.

Content moderation on social media platforms shapes the dynamics of online discourse, influencing whose voices are amplified and whose are suppressed. Recent studies have raised concerns about the fairness of content moderation practices, particularly for aggressively flagging posts from transgender and non-binary individuals as toxic. In this study, we investigate the presence of bias in harmful speech classification of gender-queer dialect online, focusing specifically on the treatment of reclaimed slurs. We introduce a novel dataset, QueerReclaimLex, based on 109 curated templates exemplifying non-derogatory uses of LGBTQ+ slurs. Dataset instances are scored by gender-queer annotators for potential harm depending on additional context about speaker identity. We systematically evaluate the performance of five off-the-shelf language models in assessing the harm of these texts and explore the effectiveness of chain-of-thought prompting to teach large language models (LLMs) to leverage author identity context. We reveal a tendency for these models to inaccurately flag texts authored by gender-queer individuals as harmful. Strikingly, across all LLMs the performance is poorest for texts that show signs of being written by individuals targeted by the featured slur (F1 <= 0.24). We highlight an urgent need for fairness and inclusivity in content moderation systems. By uncovering these biases, this work aims to inform the development of more equitable content moderation practices and contribute to the creation of inclusive online spaces for all users.

Zhivar Sourati, Farzan Karimi-Malekabadi, Meltem Ozcan et al.

Language is far more than a communication tool. A wealth of information - including but not limited to the identities, psychological states, and social contexts of its users - can be gleaned through linguistic markers, and such insights are routinely leveraged across diverse fields ranging from product development and marketing to healthcare. In four studies utilizing experimental and observational methods, we demonstrate that the widespread adoption of large language models (LLMs) as writing assistants is linked to notable declines in linguistic diversity and may interfere with the societal and psychological insights language provides. We show that while the core content of texts is retained when LLMs polish and rewrite texts, not only do they homogenize writing styles, but they also alter stylistic elements in a way that selectively amplifies certain dominant characteristics or biases while suppressing others - emphasizing conformity over individuality. By varying LLMs, prompts, classifiers, and contexts, we show that these trends are robust and consistent. Our findings highlight a wide array of risks associated with linguistic homogenization, including compromised diagnostic processes and personalization efforts, the exacerbation of existing divides and barriers to equity in settings like personnel selection where language plays a critical role in assessing candidates' qualifications, communication skills, and cultural fit, and the undermining of efforts for cultural preservation.

Abhishek Anand, Negar Mokhberian, Prathyusha Naresh Kumar et al.

Researchers have raised awareness about the harms of aggregating labels especially in subjective tasks that naturally contain disagreements among human annotators. In this work we show that models that are only provided aggregated labels show low confidence on high-disagreement data instances. While previous studies consider such instances as mislabeled, we argue that the reason the high-disagreement text instances have been hard-to-learn is that the conventional aggregated models underperform in extracting useful signals from subjective tasks. Inspired by recent studies demonstrating the effectiveness of learning from raw annotations, we investigate classifying using Multiple Ground Truth (Multi-GT) approaches. Our experiments show an improvement of confidence for the high-disagreement instances.

Bahareh Harandizadeh, Abel Salinas, Fred Morstatter

This paper explores the pressing issue of risk assessment in Large Language Models (LLMs) as they become increasingly prevalent in various applications. Focusing on how reward models, which are designed to fine-tune pretrained LLMs to align with human values, perceive and categorize different types of risks, we delve into the challenges posed by the subjective nature of preference-based training data. By utilizing the Anthropic Red-team dataset, we analyze major risk categories, including Information Hazards, Malicious Uses, and Discrimination/Hateful content. Our findings indicate that LLMs tend to consider Information Hazards less harmful, a finding confirmed by a specially developed regression model. Additionally, our analysis shows that LLMs respond less stringently to Information Hazards compared to other risks. The study further reveals a significant vulnerability of LLMs to jailbreaking attacks in Information Hazard scenarios, highlighting a critical security concern in LLM risk assessment and emphasizing the need for improved AI safety measures.

Huy Nghiem, Umang Gupta, Fred Morstatter

The propagation of offensive content through social media channels has garnered attention of the research community. Multiple works have proposed various semantically related yet subtle distinct categories of offensive speech. In this work, we explore meta-earning approaches to leverage the diversity of offensive speech corpora to enhance their reliable and efficient detection. We propose a joint embedding architecture that incorporates the input's label and definition for classification via Prototypical Network. Our model achieves at least 75% of the maximal F1-score while using less than 10% of the available training data across 4 datasets. Our experimental findings also provide a case study of training strategies valuable to combat resource scarcity.

Zhivar Sourati, Meltem Ozcan, Colin McDaniel et al.

Prior research has established associations between individuals' language usage and their personal traits; our linguistic patterns reveal information about our personalities, emotional states, and beliefs. However, with the increasing adoption of Large Language Models (LLMs) as writing assistants in everyday writing, a critical question emerges: are authors' linguistic patterns still predictive of their personal traits when LLMs are involved in the writing process? We investigate the impact of LLMs on the linguistic markers of demographic and psychological traits, specifically examining three LLMs - GPT3.5, Llama 2, and Gemini - across six different traits: gender, age, political affiliation, personality, empathy, and morality. Our findings indicate that although the use of LLMs slightly reduces the predictive power of linguistic patterns over authors' personal traits, the significant changes are infrequent, and the use of LLMs does not fully diminish the predictive power of authors' linguistic patterns over their personal traits. We also note that some theoretically established lexical-based linguistic markers lose their reliability as predictors when LLMs are used in the writing process. Our findings have important implications for the study of linguistic markers of personal traits in the age of LLMs.

Siyi Guo, Myrl G. Marmarelis, Fred Morstatter et al.

Quantifying the effects of textual interventions in social systems, such as reducing anger in social media posts to see its impact on engagement, is challenging. Real-world interventions are often infeasible, necessitating reliance on observational data. Traditional causal inference methods, typically designed for binary or discrete treatments, are inadequate for handling the complex, high-dimensional textual data. This paper addresses these challenges by proposing CausalDANN, a novel approach to estimate causal effects using text transformations facilitated by large language models (LLMs). Unlike existing methods, our approach accommodates arbitrary textual interventions and leverages text-level classifiers with domain adaptation ability to produce robust effect estimates against domain shifts, even when only the control group is observed. This flexibility in handling various text interventions is a key advancement in causal estimation for textual data, offering opportunities to better understand human behaviors and develop effective interventions within social systems.

Abha Jha, Abel Salinas, Fred Morstatter

The rise of Large Language Models (LLMs) offers transformative potential for interpreting complex legal frameworks, such as Title 18 Section 175 of the US Code, which governs biological weapons. These systems hold promise for advancing legal analysis and compliance monitoring in sensitive domains. However, this capability comes with a troubling contradiction: while LLMs can analyze and interpret laws, they also demonstrate alarming vulnerabilities in generating unsafe outputs, such as actionable steps for bioweapon creation, despite their safeguards. To address this challenge, we propose a methodology that integrates knowledge graph construction with Retrieval-Augmented Generation (RAG) to systematically evaluate LLMs' understanding of this law, their capacity to assess legal intent (mens rea), and their potential for unsafe applications. Through structured experiments, we assess their accuracy in identifying legal violations, generating prohibited instructions, and detecting unlawful intent in bioweapons-related scenarios. Our findings reveal significant limitations in LLMs' reasoning and safety mechanisms, but they also point the way forward. By combining enhanced safety protocols with more robust legal reasoning frameworks, this research lays the groundwork for developing LLMs that can ethically and securely assist in sensitive legal domains - ensuring they act as protectors of the law rather than inadvertent enablers of its violation.

Prateek Ranka, Fred Morstatter, Alexandra Graddy-Reed et al.

Classification of scientific abstracts is useful for strategic activities but challenging to automate because the sparse text provides few contextual clues. Metadata associated with the scientific publication can be used to improve performance but still often requires a semi-supervised setting. Moreover, such schemes may generate labels that lack distinction -- namely, they overlap and thus do not uniquely define the abstract. In contrast, experts label and sort these texts with ease. Here we describe an application of a process we call artificial intuition to replicate the expert's approach, using a Large Language Model (LLM) to generate metadata. We use publicly available abstracts from the United States National Science Foundation to create a set of labels, and then we test this on a set of abstracts from the Chinese National Natural Science Foundation to examine funding trends. We demonstrate the feasibility of this method for research portfolio management, technology scouting, and other strategic activities.

Maxyn Leitner, Rebecca Dorn, Fred Morstatter et al.

The recent proliferation of short form video social media sites such as TikTok has been effectively utilized for increased visibility, communication, and community connection amongst trans/nonbinary creators online. However, these same platforms have also been exploited by right-wing actors targeting trans/nonbinary people, enabling such anti-trans actors to efficiently spread hate speech and propaganda. Given these divergent groups, what are the differences in network structure between anti-trans and pro-trans communities on TikTok, and to what extent do they amplify the effects of anti-trans content? In this paper, we collect a sample of TikTok videos containing pro and anti-trans content, and develop a taxonomy of trans related sentiment to enable the classification of content on TikTok, and ultimately analyze the reply network structures of pro-trans and anti-trans communities. In order to accomplish this, we worked with hired expert data annotators from the trans/nonbinary community in order to generate a sample of highly accurately labeled data. From this subset, we utilized a novel classification pipeline leveraging Retrieval-Augmented Generation (RAG) with annotated examples and taxonomy definitions to classify content into pro-trans, anti-trans, or neutral categories. We find that incorporating our taxonomy and its logics into our classification engine results in improved ability to differentiate trans related content, and that Results from network analysis indicate many interactions between posters of pro-trans and anti-trans content exist, further demonstrating targeting of trans individuals, and demonstrating the need for better content moderation tools

Yuzhong Huang, Zhong Li, Zhang Chen et al.

In the evolving landscape of text-to-3D technology, Dreamfusion has showcased its proficiency by utilizing Score Distillation Sampling (SDS) to optimize implicit representations such as NeRF. This process is achieved through the distillation of pretrained large-scale text-to-image diffusion models. However, Dreamfusion encounters fidelity and efficiency constraints: it faces the multi-head Janus issue and exhibits a relatively slow optimization process. To circumvent these challenges, we introduce OrientDream, a camera orientation conditioned framework designed for efficient and multi-view consistent 3D generation from textual prompts. Our strategy emphasizes the implementation of an explicit camera orientation conditioned feature in the pre-training of a 2D text-to-image diffusion module. This feature effectively utilizes data from MVImgNet, an extensive external multi-view dataset, to refine and bolster its functionality. Subsequently, we utilize the pre-conditioned 2D images as a basis for optimizing a randomly initialized implicit representation (NeRF). This process is significantly expedited by a decoupled back-propagation technique, allowing for multiple updates of implicit parameters per optimization cycle. Our experiments reveal that our method not only produces high-quality NeRF models with consistent multi-view properties but also achieves an optimization speed significantly greater than existing methods, as quantified by comparative metrics.

Nikolos Gurney, Fred Morstatter, David V. Pynadath et al.

We explore the use of aggregative crowdsourced forecasting (ACF) as a mechanism to help operationalize ``collective intelligence'' of human-machine teams for coordinated actions. We adopt the definition for Collective Intelligence as: ``A property of groups that emerges from synergies among data-information-knowledge, software-hardware, and individuals (those with new insights as well as recognized authorities) that enables just-in-time knowledge for better decisions than these three elements acting alone.'' Collective Intelligence emerges from new ways of connecting humans and AI to enable decision-advantage, in part by creating and leveraging additional sources of information that might otherwise not be included. Aggregative crowdsourced forecasting (ACF) is a recent key advancement towards Collective Intelligence wherein predictions (X\% probability that Y will happen) and rationales (why I believe it is this probability that X will happen) are elicited independently from a diverse crowd, aggregated, and then used to inform higher-level decision-making. This research asks whether ACF, as a key way to enable Operational Collective Intelligence, could be brought to bear on operational scenarios (i.e., sequences of events with defined agents, components, and interactions) and decision-making, and considers whether such a capability could provide novel operational capabilities to enable new forms of decision-advantage.

Darshan Deshpande, Zhivar Sourati, Filip Ilievski et al.

Automatic assessment of the quality of arguments has been recognized as a challenging task with significant implications for misinformation and targeted speech. While real-world arguments are tightly anchored in context, existing computational methods analyze their quality in isolation, which affects their accuracy and generalizability. We propose SPARK: a novel method for scoring argument quality based on contextualization via relevant knowledge. We devise four augmentations that leverage large language models to provide feedback, infer hidden assumptions, supply a similar-quality argument, or give a counter-argument. SPARK uses a dual-encoder Transformer architecture to enable the original argument and its augmentation to be considered jointly. Our experiments in both in-domain and zero-shot setups show that SPARK consistently outperforms existing techniques across multiple metrics.

Dong-Ho Lee, Kian Ahrabian, Woojeong Jin et al.

Temporal knowledge graph (TKG) forecasting benchmarks challenge models to predict future facts using knowledge of past facts. In this paper, we apply large language models (LLMs) to these benchmarks using in-context learning (ICL). We investigate whether and to what extent LLMs can be used for TKG forecasting, especially without any fine-tuning or explicit modules for capturing structural and temporal information. For our experiments, we present a framework that converts relevant historical facts into prompts and generates ranked predictions using token probabilities. Surprisingly, we observe that LLMs, out-of-the-box, perform on par with state-of-the-art TKG models carefully designed and trained for TKG forecasting. Our extensive evaluation presents performances across several models and datasets with different characteristics, compares alternative heuristics for preparing contextual information, and contrasts to prominent TKG methods and simple frequency and recency baselines. We also discover that using numerical indices instead of entity/relation names, i.e., hiding semantic information, does not significantly affect the performance ($\pm$0.4\% Hit@1). This shows that prior semantic knowledge is unnecessary; instead, LLMs can leverage the existing patterns in the context to achieve such performance. Our analysis also reveals that ICL enables LLMs to learn irregular patterns from the historical context, going beyond simple predictions based on common or recent information.

Huy Nghiem, Fred Morstatter

Content warning: This work displays examples of explicit and/or strongly offensive language. Fueled by a surge of anti-Asian xenophobia and prejudice during the COVID-19 pandemic, many have taken to social media to express these negative sentiments. Identifying these posts is crucial for moderation and understanding the nature of hate in online spaces. In this paper, we create and annotate a corpus of tweets to explore anti-Asian hate speech with a finer level of granularity. Our analysis reveals that this emergent form of hate speech often eludes established approaches. To address this challenge, we develop a model and an accompanied efficient training regimen that incorporates agreement between annotators. Our approach produces up to 8.8% improvement in macro F1 scores over a strong established baseline, indicating its effectiveness even in settings where consensus among annotators is low. We demonstrate that we are able to identify hate speech that is systematically missed by established hate speech detectors.

Bahareh Harandizadeh, J. Hunter Priniski, Fred Morstatter

By illuminating latent structures in a corpus of text, topic models are an essential tool for categorizing, summarizing, and exploring large collections of documents. Probabilistic topic models, such as latent Dirichlet allocation (LDA), describe how words in documents are generated via a set of latent distributions called topics. Recently, the Embedded Topic Model (ETM) has extended LDA to utilize the semantic information in word embeddings to derive semantically richer topics. As LDA and its extensions are unsupervised models, they aren't defined to make efficient use of a user's prior knowledge of the domain. To this end, we propose the Keyword Assisted Embedded Topic Model (KeyETM), which equips ETM with the ability to incorporate user knowledge in the form of informative topic-level priors over the vocabulary. Using both quantitative metrics and human responses on a topic intrusion task, we demonstrate that KeyETM produces better topics than other guided, generative models in the literature.

Dong-Ho Lee, Ravi Kiran Selvam, Sheikh Muhammad Sarwar et al.

Deep neural models for named entity recognition (NER) have shown impressive results in overcoming label scarcity and generalizing to unseen entities by leveraging distant supervision and auxiliary information such as explanations. However, the costs of acquiring such additional information are generally prohibitive. In this paper, we present a novel two-stage framework (AutoTriggER) to improve NER performance by automatically generating and leveraging ``entity triggers'' which are human-readable cues in the text that help guide the model to make better decisions. Our framework leverages post-hoc explanation to generate rationales and strengthens a model's prior knowledge using an embedding interpolation technique. This approach allows models to exploit triggers to infer entity boundaries and types instead of solely memorizing the entity words themselves. Through experiments on three well-studied NER datasets, AutoTriggER shows strong label-efficiency, is capable of generalizing to unseen entities, and outperforms the RoBERTa-CRF baseline by nearly 0.5 F1 points on average.

Ninareh Mehrabi, Umang Gupta, Fred Morstatter et al.

The widespread use of Artificial Intelligence (AI) in consequential domains, such as healthcare and parole decision-making systems, has drawn intense scrutiny on the fairness of these methods. However, ensuring fairness is often insufficient as the rationale for a contentious decision needs to be audited, understood, and defended. We propose that the attention mechanism can be used to ensure fair outcomes while simultaneously providing feature attributions to account for how a decision was made. Toward this goal, we design an attention-based model that can be leveraged as an attribution framework. It can identify features responsible for both performance and fairness of the model through attention interventions and attention weight manipulation. Using this attribution framework, we then design a post-processing bias mitigation strategy and compare it with a suite of baselines. We demonstrate the versatility of our approach by conducting experiments on two distinct data types, tabular and textual.

Zaina Shaik, Filip Ilievski, Fred Morstatter

As an open and collaborative knowledge graph created by users and bots, it is possible that the knowledge in Wikidata is biased in regards to multiple factors such as gender, race, and country of citizenship. Previous work has mostly studied the representativeness of Wikidata knowledge in terms of genders of people. In this paper, we examine the race and citizenship bias in general and in regards to STEM representation for scientists, software developers, and engineers. By comparing Wikidata queries to real-world datasets, we identify the differences in representation to characterize the biases present in Wikidata. Through this analysis, we discovered that there is an overrepresentation of white individuals and those with citizenship in Europe and North America; the rest of the groups are generally underrepresented. Based on these findings, we have found and linked to Wikidata additional data about STEM scientists from the minorities. This data is ready to be inserted into Wikidata with a bot. Increasing representation of minority race and country of citizenship groups can create a more accurate portrayal of individuals in STEM.

J. Hunter Priniski, Negar Mokhberian, Bahareh Harandizadeh et al.

The viral video documenting the killing of George Floyd by Minneapolis police officer Derek Chauvin inspired nation-wide protests that brought national attention to widespread racial injustice and biased policing practices towards black communities in the United States. The use of social media by the Black Lives Matter movement was a primary route for activists to promote the cause and organize over 1,400 protests across the country. Recent research argues that moral discussions on social media are a catalyst for social change. This study sought to shed light on the moral dynamics shaping Black Lives Matter Twitter discussions by analyzing over 40,000 Tweets geo-located to Los Angeles. The goal of this study is to (1) develop computational techniques for mapping the structure of moral discourse on Twitter and (2) understand the connections between social media activism and protest.

Ninareh Mehrabi, Pei Zhou, Fred Morstatter et al.

Warning: this paper contains content that may be offensive or upsetting. Numerous natural language processing models have tried injecting commonsense by using the ConceptNet knowledge base to improve performance on different tasks. ConceptNet, however, is mostly crowdsourced from humans and may reflect human biases such as "lawyers are dishonest." It is important that these biases are not conflated with the notion of commonsense. We study this missing yet important problem by first defining and quantifying biases in ConceptNet as two types of representational harms: overgeneralization of polarized perceptions and representation disparity. We find that ConceptNet contains severe biases and disparities across four demographic categories. In addition, we analyze two downstream models that use ConceptNet as a source for commonsense knowledge and find the existence of biases in those models as well. We further propose a filtered-based bias-mitigation approach and examine its effectiveness. We show that our mitigation approach can reduce the issues in both resource and models but leads to a performance drop, leaving room for future work to build fairer and stronger commonsense models.

Ninareh Mehrabi, Muhammad Naveed, Fred Morstatter et al.

Algorithmic fairness has attracted significant attention in recent years, with many quantitative measures suggested for characterizing the fairness of different machine learning algorithms. Despite this interest, the robustness of those fairness measures with respect to an intentional adversarial attack has not been properly addressed. Indeed, most adversarial machine learning has focused on the impact of malicious attacks on the accuracy of the system, without any regard to the system's fairness. We propose new types of data poisoning attacks where an adversary intentionally targets the fairness of a system. Specifically, we propose two families of attacks that target fairness measures. In the anchoring attack, we skew the decision boundary by placing poisoned points near specific target points to bias the outcome. In the influence attack on fairness, we aim to maximize the covariance between the sensitive attributes and the decision outcome and affect the fairness of the model. We conduct extensive experiments that indicate the effectiveness of our proposed attacks.

Mehrnoosh Mirtaheri, Mohammad Rostami, Xiang Ren et al.

Most real-world knowledge graphs are characterized by a long-tail relation frequency distribution where a significant fraction of relations occurs only a handful of times. This observation has given rise to recent interest in low-shot learning methods that are able to generalize from only a few examples. The existing approaches, however, are tailored to static knowledge graphs and not easily generalized to temporal settings, where data scarcity poses even bigger problems, e.g., due to occurrence of new, previously unseen relations. We address this shortcoming by proposing a one-shot learning framework for link prediction in temporal knowledge graphs. Our proposed method employs a self-attention mechanism to effectively encode temporal interactions between entities, and a network to compute a similarity score between a given query and a (one-shot) example. Our experiments show that the proposed algorithm outperforms the state of the art baselines for two well-studied benchmarks while achieving significantly better performance for sparse relations.

Akira Matsui, Emilio Ferrara, Fred Morstatter et al.

Crowdwork often entails tackling cognitively-demanding and time-consuming tasks. Crowdsourcing can be used for complex annotation tasks, from medical imaging to geospatial data, and such data powers sensitive applications, such as health diagnostics or autonomous driving. However, the existence and prevalence of underperforming crowdworkers is well-recognized, and can pose a threat to the validity of crowdsourcing. In this study, we propose the use of a computational framework to identify clusters of underperforming workers using clickstream trajectories. We focus on crowdsourced geopolitical forecasting. The framework can reveal different types of underperformers, such as workers with forecasts whose accuracy is far from the consensus of the crowd, those who provide low-quality explanations for their forecasts, and those who simply copy-paste their forecasts from other users. Our study suggests that clickstream clustering and analysis are fundamental tools to diagnose the performance of crowdworkers in platforms leveraging the wisdom of crowds.

Ninareh Mehrabi, Yuzhong Huang, Fred Morstatter

Machine learning systems have been shown to propagate the societal errors of the past. In light of this, a wealth of research focuses on designing solutions that are "fair." Even with this abundance of work, there is no singular definition of fairness, mainly because fairness is subjective and context dependent. We propose a new fairness definition, motivated by the principle of equity, that considers existing biases in the data and attempts to make equitable decisions that account for these previous historical biases. We formalize our definition of fairness, and motivate it with its appropriate contexts. Next, we operationalize it for equitable classification. We perform multiple automatic and human evaluations to show the effectiveness of our definition and demonstrate its utility for aspects of fairness, such as the feedback loop.

Woojeong Jin, Rahul Khanna, Suji Kim et al.

Event forecasting is a challenging, yet important task, as humans seek to constantly plan for the future. Existing automated forecasting studies rely mostly on structured data, such as time-series or event-based knowledge graphs, to help predict future events. In this work, we aim to formulate a task, construct a dataset, and provide benchmarks for developing methods for event forecasting with large volumes of unstructured text data. To simulate the forecasting scenario on temporal news documents, we formulate the problem as a restricted-domain, multiple-choice, question-answering (QA) task. Unlike existing QA tasks, our task limits accessible information, and thus a model has to make a forecasting judgement. To showcase the usefulness of this task formulation, we introduce ForecastQA, a question-answering dataset consisting of 10,392 event forecasting questions, which have been collected and verified via crowdsourcing efforts. We present our experiments on ForecastQA using BERT-based models and find that our best model achieves 60.1% accuracy on the dataset, which still lags behind human performance by about 19%. We hope ForecastQA will support future research efforts in bridging this gap.

Yufei Tian, Tuhin Chakrabarty, Fred Morstatter et al.

Perspective differences exist among different cultures or languages. A lack of mutual understanding among different groups about their perspectives on specific values or events may lead to uninformed decisions or biased opinions. Automatically understanding the group perspectives can provide essential background for many downstream applications of natural language processing techniques. In this paper, we study colingual groups and use language corpora as a proxy to identify their distributional perspectives. We present a novel computational approach to learn shared understandings, and benchmark our method by building culturally-aware models for the English, Chinese, and Japanese languages. On a held out set of diverse topics including marriage, corruption, democracy, our model achieves high correlation with human judgements regarding intra-group values and inter-group differences.

Caleb Ziems, Ymir Vigfusson, Fred Morstatter

Cyberbullying is a pervasive problem in online communities. To identify cyberbullying cases in large-scale social networks, content moderators depend on machine learning classifiers for automatic cyberbullying detection. However, existing models remain unfit for real-world applications, largely due to a shortage of publicly available training data and a lack of standard criteria for assigning ground truth labels. In this study, we address the need for reliable data using an original annotation framework. Inspired by social sciences research into bullying behavior, we characterize the nuanced problem of cyberbullying using five explicit factors to represent its social and linguistic aspects. We model this behavior using social network and language-based features, which improve classifier performance. These results demonstrate the importance of representing and modeling cyberbullying as a social phenomenon.

Ninareh Mehrabi, Thamme Gowda, Fred Morstatter et al.

We study the bias in several state-of-the-art named entity recognition (NER) models---specifically, a difference in the ability to recognize male and female names as PERSON entity types. We evaluate NER models on a dataset containing 139 years of U.S. census baby names and find that relatively more female names, as opposed to male names, are not recognized as PERSON entities. We study the extent of this bias in several NER systems that are used prominently in industry and academia. In addition, we also report a bias in the datasets on which these models were trained. The result of this analysis yields a new benchmark for gender bias evaluation in named entity recognition systems. The data and code for the application of this benchmark will be publicly available for researchers to use.

Ninareh Mehrabi, Fred Morstatter, Nripsuta Saxena et al.

With the widespread use of AI systems and applications in our everyday lives, it is important to take fairness issues into consideration while designing and engineering these types of systems. Such systems can be used in many sensitive environments to make important and life-changing decisions; thus, it is crucial to ensure that the decisions do not reflect discriminatory behavior toward certain groups or populations. We have recently seen work in machine learning, natural language processing, and deep learning that addresses such challenges in different subdomains. With the commercialization of these systems, researchers are becoming aware of the biases that these applications can contain and have attempted to address them. In this survey we investigated different real-world applications that have shown biases in various ways, and we listed different sources of biases that can affect AI applications. We then created a taxonomy for fairness definitions that machine learning researchers have defined in order to avoid the existing bias in AI systems. In addition to that, we examined different domains and subdomains in AI showing what researchers have observed with regard to unfair outcomes in the state-of-the-art methods and how they have tried to address them. There are still many future directions and solutions that can be taken to mitigate the problem of bias in AI systems. We are hoping that this survey will motivate researchers to tackle these issues in the near future by observing existing work in their respective fields.

Mehrnoosh Mirtaheri, Sami Abu-El-Haija, Fred Morstatter et al.

Interest surrounding cryptocurrencies, digital or virtual currencies that are used as a medium for financial transactions, has grown tremendously in recent years. The anonymity surrounding these currencies makes investors particularly susceptible to fraud---such as "pump and dump" scams---where the goal is to artificially inflate the perceived worth of a currency, luring victims into investing before the fraudsters can sell their holdings. Because of the speed and relative anonymity offered by social platforms such as Twitter and Telegram, social media has become a preferred platform for scammers who wish to spread false hype about the cryptocurrency they are trying to pump. In this work we propose and evaluate a computational approach that can automatically identify pump and dump scams as they unfold by combining information across social media platforms. We also develop a multi-modal approach for predicting whether a particular pump attempt will succeed or not. Finally, we analyze the prevalence of bots in cryptocurrency related tweets, and observe a significant increase in bot activity during the pump attempts.

Fred Morstatter, Kai Shu, Suhang Wang et al.

Most social media platforms are largely based on text, and users often write posts to describe where they are, what they are seeing, and how they are feeling. Because written text lacks the emotional cues of spoken and face-to-face dialogue, ambiguities are common in written language. This problem is exacerbated in the short, informal nature of many social media posts. To bypass this issue, a suite of special characters called "emojis," which are small pictograms, are embedded within the text. Many emojis are small depictions of facial expressions designed to help disambiguate the emotional meaning of the text. However, a new ambiguity arises in the way that emojis are rendered. Every platform (Windows, Mac, and Android, to name a few) renders emojis according to their own style. In fact, it has been shown that some emojis can be rendered so differently that they look "happy" on some platforms, and "sad" on others. In this work, we use real-world data to verify the existence of this problem. We verify that the usage of the same emoji can be significantly different across platforms, with some emojis exhibiting different sentiment polarities on different platforms. We propose a solution to identify the intended emoji based on the platform-specific nature of the emoji used by the author of a social media post. We apply our solution to sentiment analysis, a task that can benefit from the emoji calibration technique we use in this work. We conduct experiments to evaluate the effectiveness of the mapping in this task.

Liang Wu, Fred Morstatter, Huan Liu

Sentiment in social media is increasingly considered as an important resource for customer segmentation, market understanding, and tackling other socio-economic issues. However, sentiment in social media is difficult to measure since user-generated content is usually short and informal. Although many traditional sentiment analysis methods have been proposed, identifying slang sentiment words remains untackled. One of the reasons is that slang sentiment words are not available in existing dictionaries or sentiment lexicons. To this end, we propose to build the first sentiment dictionary of slang words to aid sentiment analysis of social media content. It is laborious and time-consuming to collect and label the sentiment polarity of a comprehensive list of slang words. We present an approach to leverage web resources to construct an extensive Slang Sentiment word Dictionary (SlangSD) that is easy to maintain and extend. SlangSD is publicly available for research purposes. We empirically show the advantages of using SlangSD, the newly-built slang sentiment word dictionary for sentiment classification, and provide examples demonstrating its ease of use with an existing sentiment system.

Fred Morstatter, Nichola Lubold, Heather Pon-Barry et al.

Disaster response agencies have started to incorporate social media as a source of fast-breaking information to understand the needs of people affected by the many crises that occur around the world. These agencies look for tweets from within the region affected by the crisis to get the latest updates of the status of the affected region. However only 1% of all tweets are geotagged with explicit location information. First responders lose valuable information because they cannot assess the origin of many of the tweets they collect. In this work we seek to identify non-geotagged tweets that originate from within the crisis region. Towards this, we address three questions: (1) is there a difference between the language of tweets originating within a crisis region and tweets originating outside the region, (2) what are the linguistic patterns that can be used to differentiate within-region and outside-region tweets, and (3) for non-geotagged tweets, can we automatically identify those originating within the crisis region in real-time?