Recommender systems often operate on item catalogs clustered by genres, and user bases that have natural clusterings into user types by demographic or psychographic attributes. Prior work on system-wide diversity has mainly focused on defining intent-aware metrics among such categories and maximizing relevance of the resulting recommendations, but has not combined the notions of diversity from the two point of views of items and users. In this work, (1) we introduce two new system-wide diversity metrics to simultaneously address the problems of diversifying the categories of items that each user sees, diversifying the types of users that each item is shown, and maintaining high recommendation quality. We model this as a subgraph selection problem on the bipartite graph of candidate recommendations between users and items. (2) In the case of disjoint item categories and user types, we show that the resulting problems can be solved exactly in polynomial time, by a reduction to a minimum cost flow problem. (3) In the case of non-disjoint categories and user types, we prove NP-completeness of the objective and present efficient approximation algorithms using the submodularity of the objective. (4) Finally, we validate the effectiveness of our algorithms on the MovieLens-1m and Netflix datasets, and show that algorithms designed for our objective also perform well on sales diversity metrics, and even some intent-aware diversity metrics. Our experimental results justify the validity of our new composite diversity metrics.

Clubhouse is an audio-based social platform that launched in April 2020 and rose to popularity amidst the global COVID-19 pandemic.Unlike other platforms such as Discord, Clubhouse is entirely audio-based, and is not organized by specific communities. Following Clubhouse's surge in popularity, there has been a rise in the development of other audio-based platforms, as well as the inclusion of audio-calling features to existing platforms. In this paper, we present a framework (MIC) for analyzing audio-based social platforms that accounts for unique platform affordances, the challenges they provide to both users and moderators, and how these affordances relate to one another using MIC diagrams. Next, we demonstrate how to apply the framework to preexisting audio-based platforms and Clubhouse, highlighting key similarities and differences in affordances across these platforms. Using MIC as a lens to examine observational data from Clubhouse members we uncover user perceptions and challenges in moderating audio on the platform.

We demonstrate the advantages of using an affordance-aware framework like MIC by analyzing several social platforms over the course of two case studies. First, we analyze individual platforms using MIC and demonstrate how MIC can be used to examine the effects of platform changes on the moderation ecosystem and identify potential new challenges in moderation. Next, we use MIC to systematically compare three platforms and propose potential moderation mechanisms that each can adapt. Moderation researchers and stakeholders can use such comparisons to uncover where platforms can emulate the moderation practices of successful, established, and better-studied platforms, as well as learn from the pitfalls other platforms have encountered.

In the Priority k-Center problem, the input consists of a metric space (X, d), an integer k and for each point v ∈ X a priority radius r(v). The goal is to choose k-centers S ⊆ X to minimize max v∈X 1 r(v) d(v, S). If all r(v)'s were uniform, one obtains the classical k-center problem. Plesník [Plesník, Disc. Appl. Math. 1987] introduced this problem and gave a 2approximation algorithm matching the best possible algorithm for vanilla k-center. We show how the Priority k-Center problem is related to two different notions of fair clustering [Harris et al., NeurIPS 2018; Jung et al., FORC 2020]. Motivated by these developments we revisit the problem and, in our main technical contribution, develop a framework that yields constant factor approximation algorithms for Priority k-Center with outliers. Our framework extends to generalizations of Priority k-Center to matroid and knapsack constraints, and as a corollary, also yields algorithms with fairness guarantees in the lottery model of Harris et al.

No abstract