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.

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.

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%.

When people share updates with their friends on Facebook they have varying expectations for the feedback they will receive. In this study, we quantitatively examine the factors contributing to feedback expectations and the potential outcomes of expectation fulfillment. We conducted two sets of surveys: one asking people about their feedback expectations immediately after posting on Facebook and the other asking how the amount of feedback received on a post matched the participant's expectations. Participants were more likely to expect feedback on content they evaluated as more important, and to a lesser extent more personal. Expectations also depended on participants' age, gender, and level of activity on Facebook. When asked about feedback expectations from specific friends, participants were more likely to expect feedback from closer friends, but expectations varied considerably based on recency of communication, geographical proximity, and the type of relationship (e.g. family, co-worker). Finally, receiving more feedback relative to expectations correlated with a greater feeling of connectedness to one's Facebook friends. The findings suggest implications for the theory and the design of social network sites.

Cellular networks in developing regions continue to rely heavily on diesel for energy to provide network coverage due to the paucity of reliable grid power which directly impacts the network's economic viability and long-term sustainability. At the other extreme, solar powered cellular installations have gained prominence but have faced their own adoption challenges including inability to provide adequate and reliable 24×7 power supply, need for large land footprints and lack of efficient power storage. In this paper, we perform a detailed economic cost analysis comparing diesel powered cellular networks with solar powered cellular networks. The key goal of this paper is to establish the cross-over boundary beyond which solar powered installations are better than diesel powered alternatives. We perform a detailed analysis based on actual diesel consumption data from a large telecom operator in a developing region. Using our model, we can also easily perform an extended analysis based on future projections on solar efficiencies and future cellular network designs.

We study the complexity of rationalizing network formation. In this problem we fix an underlying model describing how selfish parties (the vertices) produce a graph by making individual decisions to form or not form incident edges. The model is equipped with a notion of stability (or equilibrium), and we observe a set of "snapshots" of graphs that are assumed to be stable. From this we would like to infer some unobserved data about the system: edge prices, or how much each vertex values short paths to each other vertex. We study two rationalization problems arising from the network formation model of Jackson and Wolinsky [14]. When the goal is to infer edge prices, we observe that the rationalization problem is easy. The problem remains easy even when rationalizing prices do not exist and we instead wish to find prices that maximize the stability of the system. In contrast, when the edge prices are given and the goal is instead to infer valuations of each vertex by each other vertex, we prove that the rationalization problem becomes NP-hard. Our proof exposes a close connection between rationalization problems and the Inequality-SAT (I-SAT) problem. Finally and most significantly, we prove that an approximation version of this NP-complete rationalization problem is NP-hard to approximate to within better than a 1/2 ratio. This shows that the trivial algorithm of setting everyone's valuations to infinity (which rationalizes all the edges present in the input graphs) or to zero (which rationalizes all the non-edges present in the input graphs) is the best possible assuming P ? NP To do this we prove a tight (1/2 + ?) -approximation hardness for a variant of I-SAT in which all coefficients are non-negative. This in turn follows from a tight hardness result for MAX-LlNR+(linear equations over the reals, with non-negative coefficients), which we prove by a (non-trivial) modification of the recent result of Guruswami and Raghavendra [10] which achieved tight hardness for this problem without the non-negativity constraint. Our technical contributions regarding the hardness of I-SAT and MAX-LINR+may be of independent interest, given the generality of these problems.