The purpose of an online electronic-payment risk detection system is to prevent leakage, i.e., the loss of revenue that occurs when users fail to pay for services or when transactions are reversed.Nonpayment prediction models are trained on datasets comprising of features available when the model is triggered and the corresponding nonpayment labels.The latter are typically only observed several weeks or even months later.Furthermore, behavior indicative of future nonpayment is highly non-stationary, and the true model may drift significantly in the gap between trigger events and label collection.To address these challenges, we use post-transaction signals to generate pseudo-labels, i.e., short-term proxies [23] or surrogate-indices [33].Our framework attains a favorable tradeoff between ameliorating bias due to drift and introducing variance due to pseudo-label noise, as demonstrated by both offline and online experiments on several nonpayment-detection systems at Meta.Our deployment on live user traffic yields a statistically significant improvement in revenue, accounting also for leakage.

We consider a recommender system that takes into account the interplay between recommendations, the evolution of user interests, and harmful content.We model the impact of recommendations on user behavior, particularly the tendency to consume harmful content.We seek recommendation policies that establish a tradeoff between maximizing click-through rate (CTR) and mitigating harm.We establish conditions under which the user profile dynamics have a stationary point, and propose algorithms for finding an optimal recommendation policy at stationarity.We experiment on a semi-synthetic movie recommendation setting initialized with real data and observe that our policies outperform baselines at simultaneously maximizing CTR and mitigating harm.

Recommender systems in social media websites provide value to their communities by recommending engaging content and meaningful connections. Scaling high-quality recommendations to billions of users in real-time requires sophisticated ranking models operating on a vast number of potential items to recommend, becoming prohibitively expensive computationally. A common technique "funnels'' these items through progressively complex models ("multi-stage''), each ranking fewer items but at higher computational cost for greater accuracy. This architecture introduces a trade-off between the cost of ranking items and providing users with the best recommendations. A key observation we make in this paper is that, all else equal, ranking more items indeed improves the overall objective but has diminishing returns. Following this observation, we provide a rigorous formulation through the framework of DR-submodularity, and argue that for a certain class of objectives (reward functions), it is possible to improve the trade-off between performance and computational cost in multi-stage ranking systems with strong theoretical guarantees. We show that this class of reward functions that provide this guarantee is large and robust to various noise models. Finally, we describe extensive experimentation of our method on three real-world recommender systems in Facebook, achieving 8.8% reduction in overall compute resources with no significant impact on recommendation quality, compared to a 0.8% quality loss in a non-personalized budget allocation.

Online platforms regularly rely on human annotators to make real-time operational decisions for tasks such as content moderation. While crowdsourcing models have been proposed for aggregating noisy labels, they do not generalize well when annotators produce a labels in a large space, e.g., generated from complex review trees. We study a novel crowdsourcing setting with D possible operational decisions or outcomes, but annotators produce labels in a larger space of size L > D which are mapped to decisions through a known mapping function. For content moderation, such labels can correspond to violation reasons (e.g. nudity, violence), while the space of decisions is binary: remove the content or keep it up. In this setting, it is more important to make the right decision rather than estimating the correct underlying label. Existing methods typically separate out the labels to decisions mapping from the modeling of annotators, leading to sub-optimal statistical inference efficiency and excessive computation complexity. We propose a novel confusion matrix model for each annotator that leverages this mapping. Our model is parameterized in a hierarchical manner with both population parameters shared across annotators to model shared confusions and individual parameters to admit heterogeneity among annotators. With extensive numerical experiments, we demonstrate that the proposed model substantially improves accuracy over existing methods and scales well for moderate and large L. In a real-world application on content moderation at Meta, the proposed method offers a 13% improvement in AUC over prior methods, including Meta's existing model in production.

Social media platforms like Facebook and YouTube connect people with communities that reflect their own values and experiences. People discover new communities either organically or through algorithmic recommendations based on their interests and preferences. We study online journeys users take through these communities, focusing particularly on ones that may lead to problematic outcomes. In particular, we propose and explore the concept of gateways, namely, entities associated with a higher likelihood of subsequent engagement with problematic content. We show, via a real-world application on Facebook groups, that a simple definition of gateway entities can be leveraged to reduce exposure to problematic content by 1% without any adverse impact on user engagement metrics. Motivated by this finding, we propose several formal definitions of gateways, via both frequentist and survival analysis methods, and evaluate their efficacy in predicting user behavior through offline experiments. Frequentist, duration-insensitive methods predict future harmful engagements with an 0.64–0.83 AUC, while survival analysis methods improve this to 0.72–0.90 AUC.

We study a crowdsourcing setting where we need to infer the latent truth about a task given observed labels together with context in the form of a classifier score. We present Theodon, a hierarchical non-parametric Bayesian model, developed and deployed at Meta, that captures both the prevalence of label categories and the accuracy of labelers as functions of the classifier score. Theodon uses Gaussian processes to model the non-uniformity of mistakes over the range of classifier scores. For our experiments, we used data generated from integrity applications at Meta as well as public datasets. We showed that Theodon (1) obtains 1-4% improvement in AUC-PR predictions on items' true labels compared to state-of-the-art baselines for public datasets, (2) is effective as a calibration method, and (3) provides detailed insights on labelers' performances.

Recommender systems have become increasingly accurate in suggesting content to users, resulting in users primarily consuming content through recommendations. This can cause the user's interest to narrow toward the recommended content, something we refer to as preference amplification. While this can contribute to increased engagement, it can also lead to negative experiences such as lack of diversity and echo chambers. We propose a theoretical framework for studying such amplification in a matrix factorization based recommender system. We model the dynamics of the system, where users interact with the recommender systems and gradually "drift'' toward the recommended content, with the recommender system adapting, based on user feedback, to the updated preferences. We study the conditions under which preference amplification manifests, and validate our results with simulations. Finally, we evaluate mitigation strategies that prevent the adverse effects of preference amplification and present experimental results using a real-world large-scale video recommender system showing that by reducing exposure to potentially objectionable content we can increase user engagement by up to 2%.

Large online services employ thousands of people to label content for applications such as video understanding, natural language processing, and content policy enforcement. While labelers typically reach their decisions by following a well-defined "protocol'', humans may still make mistakes. A common countermeasure is to have multiple people review the same content; however, this process is often time-intensive and requires accurate aggregation of potentially noisy decisions.

In this paper, we study the problem of early detection of fake user accounts on social networks based solely on their network connectivity with other users. Removing such accounts is a core task for maintaining the integrity of social networks, and early detection helps to reduce the harm that such accounts inflict. However, new fake accounts are notoriously difficult to detect via graph-based algorithms, as their small number of connections are unlikely to reflect a significant structural difference from those of new real accounts. We present the SybilEdge algorithm, which determines whether a new user is a fake account (‘sybil’) by aggregating over (I) her choices of friend request targets and (II) these targets’ respective responses. SybilEdge performs this aggregation giving more weight to a user’s choices of targets to the extent that these targets are preferred by other fakes versus real users, and also to the extent that these targets respond differently to fakes versus real users. We show that SybilEdge rapidly detects new fake users at scale on the Facebook network and outperforms state-of-the-art algorithms. We also show that SybilEdge is robust to label noise in the training data, to different prevalences of fake accounts in the network, and to several different ways fakes can select targets for their friend requests. To our knowledge, this is the first time a graph-based algorithm has been shown to achieve high performance (AUC > 0.9) on new users who have only sent a small number of friend requests.

How can we design a product or movie that will attract, for example, the interest of Pennsylvania adolescents or liberal newspaper critics? What should be the genre of that movie and who should be in the cast? In this work, we seek to identify how we can design new movies with features tailored to a specific user population. We formulate the movie design as an optimization problem over the inference of user-feature scores and selection of the features that maximize the number of attracted users. Our approach, PNP, is based on a heterogeneous, tripartite graph of users, movies, and features (e.g. actors, directors, genres), where users rate movies and features contribute to movies. We learn the preferences by leveraging user similarities defined through different types of relations, and show that our method outperforms state-of-the-art approaches, including matrix factorization and other heterogeneous graph-based analysis. We evaluate PNP on publicly available real-world data and show that it is highly scalable and effectively provides movie designs oriented towards different groups of users, including men, women, and adolescents.

Malvertising is a malicious activity that leverages advertising to distribute various forms of malware. Because advertising is the key revenue generator for numerous Internet companies, large ad networks, such as Google, Yahoo and Microsoft, invest a lot of effort to mitigate malicious ads from their ad networks. This drives adversaries to look for alternative methods to deploy malvertising. In this paper, we show that browser extensions that use ads as their monetization strategy often facilitate the deployment of malvertising. Moreover, while some extensions simply serve ads from ad networks that support malvertising, other extensions maliciously alter the content of visited webpages to force users into installing malware. To measure the extent of these behaviors we developed Expector, a system that automatically inspects and identifies browser extensions that inject ads, and then classifies these ads as malicious or benign based on their landing pages. Using Expector, we automatically inspected over 18,000 Chrome browser extensions. We found 292 extensions that inject ads, and detected 56 extensions that participate in malvertising using 16 different ad networks and with a total user base of 602,417.

Recommender systems leverage user demographic information, such as age, gender, etc., to personalize recommendations and better place their targeted ads. Oftentimes, users do not volunteer this information due to privacy concerns, or due to a lack of initiative in filling out their online profiles. We illustrate a new threat in which a recommender learns private attributes of users who do not voluntarily disclose them. We design both passive and active attacks that solicit ratings for strategically selected items, and could thus be used by a recommender system to pursue this hidden agenda. Our methods are based on a novel usage of Bayesian matrix factorization in an active learning setting. Evaluations on multiple datasets illustrate that such attacks are indeed feasible and use significantly fewer rated items than static inference methods. Importantly, they succeed without sacrificing the quality of recommendations to users.

User demographics, such as age, gender and ethnicity, are routinely used for targeting content and advertising products to users. Similarly, recommender systems utilize user demographics for personalizing recommendations and overcoming the cold-start problem. Often, privacy-concerned users do not provide these details in their online profiles. In this work, we show that a recommender system can infer the gender of a user with high accuracy, based solely on the ratings provided by users (without additional metadata), and a relatively small number of users who share their demographics. Focusing on gender, we design techniques for effectively adding ratings to a user's profile for obfuscating the user's gender, while having an insignificant effect on the recommendations provided to that user.

Peer-to-Peer (p2p) networks are used for sharing content by millions of users. Often, meta-data used for searching is missing or wrong, making it difficult for users to find content. Moreover, searching for new content is almost impossible. Recommender systems are unable to handle p2p data due to inherent difficulties, such as implicit ranking, noise and the extreme dimensions and sparseness of the network.