Aron Culotta

Karthik Shivaram, Mustafa Bilgic, Matthew Shapiro et al.

We propose novel methods to identify tweets that criticize partisan news sources. Prior work suggests that criticism, ridicule, and distrust of news media all play important roles in hyperpartisanship, misinformation, and filter bubble formation. Thus, understanding the prevalence and temporal dynamics of media-targeted criticism can provide us with updated tools to assess the health of the information ecosystem. There is a scarcity of labeled data for this task, and we develop a weakly supervised learning approach that leverages multiple noisy labeling functions based on both the content of the tweet as well as the historical news sharing behavior of the user. Using this classifier, we explore how tweets expressing criticism vary by user, news source, and time, finding substantial spikes in media criticism during politically polarizing events, such as the investigation into Russian interference in the 2016 U.S.~elections and the 2017 ``unite the right'' rally in Charlottesville. This type of media-targeting criticism is also more likely to occur after users have been exposed to unreliable and hyperpartisan media.

Karthik Shivaram, Mustafa Bilgic, Matthew Shapiro et al.

Understanding how political news consumption changes over time can provide insights into issues such as hyperpartisanship, filter bubbles, and misinformation. To investigate long-term trends of news consumption, we curate a collection of over 60M tweets from politically engaged users over seven years, annotating ~10% with mentions of news outlets and their political leaning. We then train a neural network to forecast the political lean of news articles Twitter users will engage with, considering both past news engagements as well as tweet content. Using the learned representation of this model, we cluster users to discover salient patterns of long-term news engagement. Our findings include the following: (1) hyperpartisan users are more engaged with news; (2) right-leaning users engage with contra-partisan sources more than left-leaning users; (3) topics such as immigration, COVID-19, Islamaphobia, and gun control are salient indicators of engagement with low quality news sources.

Linsen Li, Aron Culotta, Nicholas Mattei

Online reviews provide valuable insights into the perceived quality of facets of a product or service. While aspect-based sentiment analysis has focused on extracting these facets from reviews, there is less work understanding the impact of each aspect on overall perception. This is particularly challenging given correlations among aspects, making it difficult to isolate the effects of each. This paper introduces a methodology based on recent advances in text-based causal analysis, specifically CausalBERT, to disentangle the effect of each factor on overall review ratings. We enhance CausalBERT with three key improvements: temperature scaling for better calibrated treatment assignment estimates; hyperparameter optimization to reduce confound overadjustment; and interpretability methods to characterize discovered confounds. In this work, we treat the textual mentions in reviews as proxies for real-world attributes. We validate our approach on real and semi-synthetic data from over 600K reviews of U.S. K-12 schools. We find that the proposed enhancements result in more reliable estimates, and that perception of school administration and performance on benchmarks are significant drivers of overall school ratings.

Disa Sariola, Patrick Button, Aron Culotta et al.

Artificial intelligence systems, especially those using machine learning, are being deployed in domains from hiring to loan issuance in order to automate these complex decisions. Judging both the effectiveness and fairness of these AI systems, and their human decision making counterpart, is a complex and important topic studied across both computational and social sciences. Within machine learning, a common way to address bias in downstream classifiers is to resample the training data to offset disparities. For example, if hiring rates vary by some protected class, then one may equalize the rate within the training set to alleviate bias in the resulting classifier. While simple and seemingly effective, these methods have typically only been evaluated using data obtained through convenience samples, introducing selection bias and label bias into metrics. Within the social sciences, psychology, public health, and medicine, audit studies, in which fictitious ``testers'' (e.g., resumes, emails, patient actors) are sent to subjects (e.g., job openings, businesses, doctors) in randomized control trials, provide high quality data that support rigorous estimates of discrimination. In this paper, we investigate how data from audit studies can be used to improve our ability to both train and evaluate automated hiring algorithms. We find that such data reveals cases where the common fairness intervention method of equalizing base rates across classes appears to achieve parity using traditional measures, but in fact has roughly 10% disparity when measured appropriately. We additionally introduce interventions based on individual treatment effect estimation methods that further reduce algorithmic discrimination using this data.

Zhao Wang, Kai Shu, Aron Culotta

Recent work has raised concerns on the risk of spurious correlations and unintended biases in statistical machine learning models that threaten model robustness and fairness. In this paper, we propose a simple and intuitive regularization approach to integrate causal knowledge during model training and build a robust and fair model by emphasizing causal features and de-emphasizing spurious features. Specifically, we first manually identify causal and spurious features with principles inspired from the counterfactual framework of causal inference. Then, we propose a regularization approach to penalize causal and spurious features separately. By adjusting the strength of the penalty for each type of feature, we build a predictive model that relies more on causal features and less on non-causal features. We conduct experiments to evaluate model robustness and fairness on three datasets with multiple metrics. Empirical results show that the new models built with causal awareness significantly improve model robustness with respect to counterfactual texts and model fairness with respect to sensitive attributes.

Zhao Wang, Aron Culotta

Spurious correlations threaten the validity of statistical classifiers. While model accuracy may appear high when the test data is from the same distribution as the training data, it can quickly degrade when the test distribution changes. For example, it has been shown that classifiers perform poorly when humans make minor modifications to change the label of an example. One solution to increase model reliability and generalizability is to identify causal associations between features and classes. In this paper, we propose to train a robust text classifier by augmenting the training data with automatically generated counterfactual data. We first identify likely causal features using a statistical matching approach. Next, we generate counterfactual samples for the original training data by substituting causal features with their antonyms and then assigning opposite labels to the counterfactual samples. Finally, we combine the original data and counterfactual data to train a robust classifier. Experiments on two classification tasks show that a traditional classifier trained on the original data does very poorly on human-generated counterfactual samples (e.g., 10%-37% drop in accuracy). However, the classifier trained on the combined data is more robust and performs well on both the original test data and the counterfactual test data (e.g., 12%-25% increase in accuracy compared with the traditional classifier). Detailed analysis shows that the robust classifier makes meaningful and trustworthy predictions by emphasizing causal features and de-emphasizing non-causal features.

Zhao Wang, Jennifer Cutler, Aron Culotta

Public entities such as companies and politicians increasingly use online social networks to communicate directly with their constituencies. Often, this public messaging is aimed at aligning the entity with a particular cause or issue, such as the environment or public health. However, as a consumer or voter, it can be difficult to assess an entity's true commitment to a cause based on public messaging. In this paper, we present a text classification approach to categorize a message according to its commitment level toward a cause. We then compare the volume of such messages with external ratings based on entities' actions (e.g., a politician's voting record with respect to the environment or a company's rating from environmental non-profits). We find that by distinguishing between low- and high- level commitment messages, we can more reliably identify truly committed entities. Furthermore, by measuring the discrepancy between classified messages and external ratings, we can identify entities whose public messaging does not align with their actions, thereby providing a methodology to identify potentially "inauthentic" messaging campaigns.

Zhao Wang, Aron Culotta

The predictions of text classifiers are often driven by spurious correlations -- e.g., the term `Spielberg' correlates with positively reviewed movies, even though the term itself does not semantically convey a positive sentiment. In this paper, we propose a method to distinguish spurious and genuine correlations in text classification. We treat this as a supervised classification problem, using features derived from treatment effect estimators to distinguish spurious correlations from "genuine" ones. Due to the generic nature of these features and their small dimensionality, we find that the approach works well even with limited training examples, and that it is possible to transport the word classifier to new domains. Experiments on four datasets (sentiment classification and toxicity detection) suggest that using this approach to inform feature selection also leads to more robust classification, as measured by improved worst-case accuracy on the samples affected by spurious correlations.

Zhao Wang, Anna Sapienza, Aron Culotta et al.

Both offline and online human behaviors are affected by personality. Of special interests are online games, where players have to impersonate specific roles and their behaviors are extensively tracked by the game. In this paper, we propose to study the relationship between players' personality and game behavior in League of Legends (LoL), one of the most popular Multiplayer Online Battle Arena (MOBA) games. We use linear mixed effects (LME) models to describe relationships between players' personality traits (measured by the Five Factor Model) and two major aspects of the game: the impersonated roles and in-game actions. On the one hand, we study relationships within the game environment by modeling role attributes from match behaviors and vice versa. On the other hand, we analyze the relationship between a player's five personality traits and their game behavior by showing significant correlations between each personality trait and the set of corresponding behaviors. Our findings suggest that personality and behavior are highly entangled and provide a new perspective to understand how personality can affect behavior in role-based online games.

Zhao Wang, Aron Culotta

Studies across many disciplines have shown that lexical choice can affect audience perception. For example, how users describe themselves in a social media profile can affect their perceived socio-economic status. However, we lack general methods for estimating the causal effect of lexical choice on the perception of a specific sentence. While randomized controlled trials may provide good estimates, they do not scale to the potentially millions of comparisons necessary to consider all lexical choices. Instead, in this paper, we first offer two classes of methods to estimate the effect on perception of changing one word to another in a given sentence. The first class of algorithms builds upon quasi-experimental designs to estimate individual treatment effects from observational data. The second class treats treatment effect estimation as a classification problem. We conduct experiments with three data sources (Yelp, Twitter, and Airbnb), finding that the algorithmic estimates align well with those produced by randomized-control trials. Additionally, we find that it is possible to transfer treatment effect classifiers across domains and still maintain high accuracy.

Ping Liu, Joshua Guberman, Libby Hemphill et al.

Online antisocial behavior, such as cyberbullying, harassment, and trolling, is a widespread problem that threatens free discussion and has negative physical and mental health consequences for victims and communities. While prior work has proposed automated methods to identify hostile comments in online discussions, these methods work retrospectively on comments that have already been posted, making it difficult to intervene before an interaction escalates. In this paper we instead consider the problem of forecasting future hostilities in online discussions, which we decompose into two tasks: (1) given an initial sequence of non-hostile comments in a discussion, predict whether some future comment will contain hostility; and (2) given the first hostile comment in a discussion, predict whether this will lead to an escalation of hostility in subsequent comments. Thus, we aim to forecast both the presence and intensity of hostile comments based on linguistic and social features from earlier comments. To evaluate our approach, we introduce a corpus of over 30K annotated Instagram comments from over 1,100 posts. Our approach is able to predict the appearance of a hostile comment on an Instagram post ten or more hours in the future with an AUC of .82 (task 1), and can furthermore distinguish between high and low levels of future hostility with an AUC of .91 (task 2).

Ehsan Mohammady Ardehaly, Aron Culotta

Deep learning algorithms have recently produced state-of-the-art accuracy in many classification tasks, but this success is typically dependent on access to many annotated training examples. For domains without such data, an attractive alternative is to train models with light, or distant supervision. In this paper, we introduce a deep neural network for the Learning from Label Proportion (LLP) setting, in which the training data consist of bags of unlabeled instances with associated label distributions for each bag. We introduce a new regularization layer, Batch Averager, that can be appended to the last layer of any deep neural network to convert it from supervised learning to LLP. This layer can be implemented readily with existing deep learning packages. To further support domains in which the data consist of two conditionally independent feature views (e.g. image and text), we propose a co-training algorithm that iteratively generates pseudo bags and refits the deep LLP model to improve classification accuracy. We demonstrate our models on demographic attribute classification (gender and race/ethnicity), which has many applications in social media analysis, public health, and marketing. We conduct experiments to predict demographics of Twitter users based on their tweets and profile image, without requiring any user-level annotations for training. We find that the deep LLP approach outperforms baselines for both text and image features separately. Additionally, we find that co-training algorithm improves image and text classification by 4% and 8% absolute F1, respectively. Finally, an ensemble of text and image classifiers further improves the absolute F1 measure by 4% on average.

Virgile Landeiro, Aron Culotta

As statistical classifiers become integrated into real-world applications, it is important to consider not only their accuracy but also their robustness to changes in the data distribution. In this paper, we consider the case where there is an unobserved confounding variable $z$ that influences both the features $\mathbf{x}$ and the class variable $y$. When the influence of $z$ changes from training to testing data, we find that the classifier accuracy can degrade rapidly. In our approach, we assume that we can predict the value of $z$ at training time with some error. The prediction for $z$ is then fed to Pearl's back-door adjustment to build our model. Because of the attenuation bias caused by measurement error in $z$, standard approaches to controlling for $z$ are ineffective. In response, we propose a method to properly control for the influence of $z$ by first estimating its relationship with the class variable $y$, then updating predictions for $z$ to match that estimated relationship. By adjusting the influence of $z$, we show that we can build a model that exceeds competing baselines on accuracy as well as on robustness over a range of confounding relationships.

Shreesh Kumara Bhat, Aron Culotta

Consumer protection agencies are charged with safeguarding the public from hazardous products, but the thousands of products under their jurisdiction make it challenging to identify and respond to consumer complaints quickly. From the consumer's perspective, online reviews can provide evidence of product defects, but manually sifting through hundreds of reviews is not always feasible. In this paper, we propose a system to mine Amazon.com reviews to identify products that may pose safety or health hazards. Since labeled data for this task are scarce, our approach combines positive unlabeled learning with domain adaptation to train a classifier from consumer complaints submitted to the U.S. Consumer Product Safety Commission. On a validation set of manually annotated Amazon product reviews, we find that our approach results in an absolute F1 score improvement of 8% over the best competing baseline. Furthermore, we apply the classifier to Amazon reviews of known recalled products; the classifier identifies reviews reporting safety hazards prior to the recall date for 45% of the products. This suggests that the system may be able to provide an early warning system to alert consumers to hazardous products before an official recall is announced.

Reid Priedhorsky, Aron Culotta, Sara Y. Del Valle

Social Internet content plays an increasingly critical role in many domains, including public health, disaster management, and politics. However, its utility is limited by missing geographic information; for example, fewer than 1.6% of Twitter messages (tweets) contain a geotag. We propose a scalable, content-based approach to estimate the location of tweets using a novel yet simple variant of gaussian mixture models. Further, because real-world applications depend on quantified uncertainty for such estimates, we propose novel metrics of accuracy, precision, and calibration, and we evaluate our approach accordingly. Experiments on 13 million global, comprehensively multi-lingual tweets show that our approach yields reliable, well-calibrated results competitive with previous computationally intensive methods. We also show that a relatively small number of training data are required for good estimates (roughly 30,000 tweets) and models are quite time-invariant (effective on tweets many weeks newer than the training set). Finally, we show that toponyms and languages with small geographic footprint provide the most useful location signals.