arXiv subjects

Fred Morstatter

Publications and source records attributed to Fred Morstatter.

At least 55 records · Page 3Linked to original sources

arXiv2023

Non-Binary Gender Expression in Online Interactions

Many openly non-binary gender individuals participate in social networks. However, the relationship between gender and online interactions is not well understood, which may result in disparate treatment by large language models. We investigate individual identity on Twitter, focusing on gender expression as represented by users chosen pronouns. We find that non-binary groups tend to receive less attention in the form of likes and followers. We also find that nonbinary users send and receive tweets with above-average toxicity. The study highlights the importance of considering gender as a spectrum, rather than a binary, in understanding online interactions and expression.

cs.SI↗

arXiv2023

Retweets Amplify the Echo Chamber Effect

The growing prominence of social media in public discourse has led to a greater scrutiny of the quality of online information and the role it plays in amplifying political polarization. However, studies of polarization on social media platforms like Twitter have been hampered by the difficulty of collecting data about the social graph, specifically follow links that shape the echo chambers users join as well as what they see in their timelines. As a proxy of the follower graph, researchers use retweets, although it is not clear how this choice affects analysis. Using a sample of the Twitter follower graph and the tweets posted by users within it, we reconstruct the retweet graph and quantify its impact on the measures of echo chambers and exposure. While we find that echo chambers exist in both graphs, they are more pronounced in the retweet graph. We compare the information users see via their follower and retweet networks to show that retweeted accounts share systematically more polarized content. This bias cannot be explained by the activity or polarization within users' own follower graph neighborhoods but by the increased attention they pay to accounts that are ideologically aligned with their own views. Our results suggest that studies relying on the retweet graphs overestimate the echo chamber effects and exposure to polarized information.

cs.SI↗

arXiv2023

Modeling Cross-Cultural Pragmatic Inference with Codenames Duet

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.

cs.CL↗

arXiv2023

AutoTriggER: Label-Efficient and Robust Named Entity Recognition with Auxiliary Trigger Extraction

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.

cs.CL↗

arXiv2023

Just Another Day on Twitter: A Complete 24 Hours of Twitter Data

At the end of October 2022, Elon Musk concluded his acquisition of Twitter. In the weeks and months before that, several questions were publicly discussed that were not only of interest to the platform's future buyers, but also of high relevance to the Computational Social Science research community. For example, how many active users does the platform have? What percentage of accounts on the site are bots? And, what are the dominating topics and sub-topical spheres on the platform? In a globally coordinated effort of 80 scholars to shed light on these questions, and to offer a dataset that will equip other researchers to do the same, we have collected all 375 million tweets published within a 24-hour time period starting on September 21, 2022. To the best of our knowledge, this is the first complete 24-hour Twitter dataset that is available for the research community. With it, the present work aims to accomplish two goals. First, we seek to answer the aforementioned questions and provide descriptive metrics about Twitter that can serve as references for other researchers. Second, we create a baseline dataset for future research that can be used to study the potential impact of the platform's ownership change.

cs.SI↗

arXiv2023

Gender and Prestige Bias in Coronavirus News Reporting

Journalists play a vital role in surfacing issues of societal importance, but their choices of what to highlight and who to interview are influenced by societal biases. In this work, we use natural language processing tools to measure these biases in a large corpus of news articles about the Covid-19 pandemic. Specifically, we identify when experts are quoted in news and extract their names and institutional affiliations. We enrich the data by classifying each expert's gender, the type of organization they belong to, and for academic institutions, their ranking. Our analysis reveals disparities in the representation of experts in news. We find a substantial gender gap, where men are quoted three times more than women. The gender gap varies by partisanship of the news source, with conservative media exhibiting greater gender bias. We also identify academic prestige bias, where journalists turn to experts from highly-ranked academic institutions more than experts from less prestigious institutions, even if the latter group has more public health expertise. Liberal news sources exhibit slightly more prestige bias than conservative sources. Equality of representation is essential to enable voices from all groups to be heard. By auditing bias, our methods help identify blind spots in news coverage.

cs.SI↗

arXiv2022

Noise Audits Improve Moral Foundation Classification

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.

cs.CL↗

arXiv2022

"Stop Asian Hate!" : Refining Detection of Anti-Asian Hate Speech During the COVID-19 Pandemic

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.

cs.CL↗

arXiv2022

Robust Conversational Agents against Imperceptible Toxicity Triggers

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.

cs.CL↗

arXiv2022

Anchor Attention for Hybrid Crowd Forecasts Aggregation

In a crowd forecasting system, aggregation is an algorithm that returns aggregated probabilities for each question based on the probabilities provided per question by each individual in the crowd. Various aggregation methods have been proposed, but simple strategies like linear averaging or selecting the best-performing individual remain competitive. With the recent advance in neural networks, we model forecasts aggregation as a machine translation task, that translates from a sequence of individual forecasts into aggregated forecasts, based on proposed Anchor Attention between questions and forecasters. We evaluate our approach using data collected on our forecasting platform and publicly available Good Judgement Project dataset, and show that our method outperforms current state-of-the-art aggregation approaches by learning a good representation of forecaster and question.

stat.AP↗

arXiv2022

Partisan Asymmetries in Exposure to Misinformation

Health misinformation is believed to have contributed to vaccine hesitancy during the Covid-19 pandemic, highlighting concerns about the role of social media in polarization and social stability. While previous research has identified a link between political partisanship and misinformation sharing online, the interaction between partisanship and how much misinformation people see within their social networks has not been well studied. As a result, we do not know whether partisanship drives exposure to misinformation or people selectively share misinformation despite being exposed to factual content. We study Twitter discussions about the Covid-19 pandemic, classifying users ideologically along political and factual dimensions. We find partisan asymmetries in both sharing behaviors and exposure, with conservatives more likely to see and share misinformation and moderate liberals seeing the most factual content. We identify multi-dimensional echo chambers that expose users to ideologically congruent content; however, the interaction between political and factual dimensions creates conditions for the highly polarized users -- hardline conservatives and liberals -- to amplify misinformation. Despite this, misinformation receives less attention than factual content and political moderates, who represent the bulk of users in our sample, help filter out misinformation, reducing the amount of low factuality content in the information ecosystem. Identifying the extent of polarization and how political ideology can exacerbate misinformation can potentially help public health experts and policy makers improve their messaging to promote consensus.

cs.SI↗

arXiv2022

A Survey on Bias and Fairness in Machine Learning

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.

cs.LG↗

arXiv2021

Keyword Assisted Embedded Topic Model

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.

cs.IR↗

arXiv2021

Lawyers are Dishonest? Quantifying Representational Harms in Commonsense Knowledge Resources

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.

cs.CL↗

arXiv2021

Attributing Fair Decisions with Attention Interventions

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.

cs.AI↗

arXiv2021

Mapping Moral Valence of Tweets Following the Killing of George Floyd

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.

cs.SI↗

arXiv2021

Analyzing Race and Country of Citizenship Bias in Wikidata

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.

cs.AI↗

arXiv2021

ForecastQA: A Question Answering Challenge for Event Forecasting with Temporal Text Data

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.

cs.LG↗