Algorithms personalize social media feeds by ranking posts from the inventory of a user's network. However, the combination of network structure and user activity can distort the perceived popularity of user trait...
ISBN:
(纸本)9783031785375;9783031785382
Algorithms personalize social media feeds by ranking posts from the inventory of a user's network. However, the combination of network structure and user activity can distort the perceived popularity of user traits in the network well before any personalization step. To measure this "exposure bias" and how users might perceive their network when subjected to personalization, we conducted an analysis using archival X (formerly Twitter) data with a fixed inventory. We compare different ways recommender systems rank-order feeds: by recency, by popularity, based one the expected probability of engagement, and random sorting. Our results suggest that users who are subject to simpler algorithmic feeds experience significantly higher exposure bias compared to those with chronologically-sorted, popularity-sorted and deep-learning recommender models. Furthermore, we identify two key factors for bias mitigation: the effective degree-attribute correlation and session length. These factors can be adjusted to control the level of exposure bias experienced by users. To conclude we describe how this framework can extend to other platforms. Our findings highlight how the interactions between social networks and algorithmic curation shape-and distortuser's online experience.
The goal of this paper is to rigorously interrogate conventional wisdom about centralization in block-building (due to, e.g., MEV and private order flow) and the outsourcing of block-building by validators to speciali...
ISBN:
(纸本)9783031786754;9783031786761
The goal of this paper is to rigorously interrogate conventional wisdom about centralization in block-building (due to, e.g., MEV and private order flow) and the outsourcing of block-building by validators to specialists (i.e., proposer-builder separation): 1. Does heterogeneity in skills and knowledge across block producers inevitably lead to centralization? 2. Does proposer-builder separation eliminate heterogeneity and preserve decentralization among proposers? This paper develops mathematical models and results that offer answers to these questions: 1. In a game-theoretic model with endogenous staking, heterogeneous block producer rewards, and staking costs, we quantify the extent to which heterogeneous rewards lead to concentration in the equilibrium staking distribution. 2. In a stochastic model in which heterogeneous block producers repeatedly reinvest rewards into staking, we quantify, as a function of the block producer heterogeneity, the rate at which stake concentrates on the most sophisticated block producers. 3. In a model with heterogeneous proposers and specialized builders, we quantify, as a function of the competitiveness of the builder ecosystem, the extent to which proposer-builder separation reduces the heterogeneity in rewards across different proposers. Our models and results take advantage of connections to contest design, Polya urn processes, and auction theory.
Sex trafficking is an international crisis that has permeated the online realm, enabling traffickers to build a larger client base and evade law enforcement. The current study was conducted to identify sex trafficking...
ISBN:
(纸本)9783031785535;9783031785542
Sex trafficking is an international crisis that has permeated the online realm, enabling traffickers to build a larger client base and evade law enforcement. The current study was conducted to identify sex trafficking by comparing advertisements (ads) with matching contact information. Ads with matching contact information and different descriptions of individuals may signify that one individual is controlling the sale of multiple people, potentially indicating trafficking. A customized web scraper was utilized to collect information from ads from the 'personals' category of ***;those with social media handles, phone/WhatsApp numbers, email addresses, and 'click to view' numbers were extracted. Ads were grouped based on matching contact information and qualitatively analyzed;groups with different physical descriptions of girls were coded as potential trafficking. Ads with matching contacts and listings of multiple girls were searched with sex trafficking keywords and coded as potential trafficking or agency if at least half the ads contained over three keywords. From 7,328 unique contacts, 97 were coded as potential trafficking and 67 as potential trafficking or agency. The practical implication of this project is that an automated tool may be developed that uses these methods to identify trafficking advertisements to take them down, potentially aiding victims.
How can we effectively model arguments communicated in diverse environments? On the one hand, there is a great opportunity with the abundance of digitized speech across different contexts including online forums, offi...
ISBN:
(纸本)9783031785405;9783031785412
How can we effectively model arguments communicated in diverse environments? On the one hand, there is a great opportunity with the abundance of digitized speech across different contexts including online forums, official proceedings, or transcripts of spoken debates. On the other hand, there is a great challenge in correctly detecting arguments, especially since each medium has its own set of conventions, lingo, affordances, and styles of argumentative engagement. We propose WIBA, a novel framework and suite of methods that enable the comprehensive understanding of " What Is Being Argued" across contexts. Our approach develops a comprehensive framework that detects: (a) the existence, (b) the topic, and (c) the stance of an argument, correctly accounting for the logical dependence among the three tasks. Our algorithm leverages the fine-tuning and prompt-engineering of Large Language Models. We evaluate our approach and show that it performs well in all the three capabilities. First, we develop and release an Argument Detection model that can classify a piece of text as an argument with an F-1 score between 79% and 86% on three different benchmark datasets. Second, we release a language model that can identify the topic being argued in a sentence, be it implicit or explicit, with an average similarity score of 71%, outperforming current naive methods by nearly 40%. Finally, we develop a method for Argument Stance Classification, and evaluate the capability of our approach, showing it achieves a classification F-1 score between 71% and 78% across three diverse benchmark datasets. Our evaluation demonstrates that WIBA allows the comprehensive understanding of What Is Being Argued in large corpora across diverse contexts, which is of core interest to many applications in linguistics, communication, and social and computerscience. To facilitate accessibility to the advancements outlined in this work, we release WIBA as a free open access platform (***) and API.
Graph neural networks (GNNs) have demonstrated remarkable success in addressing a variety of node classification problems. Cross-network node classification (CNNC) extends the GNN formulation to a multi-network settin...
ISBN:
(纸本)9783031785375;9783031785382
Graph neural networks (GNNs) have demonstrated remarkable success in addressing a variety of node classification problems. Cross-network node classification (CNNC) extends the GNN formulation to a multi-network setting, enabling the classification to be performed on an unlabeled target network. However, applying GNNs to a multi-network setting in practice is a challenge due to the possible presence of concept drift and the need to account for link biases in the graph data. In this paper we present FOCI, a powerful, model-agnostic approach for cross-network node classification that enables the GNN to overcome the concept drift issue while mitigating potential biases in the data. FOCI utilizes a fair Sinkhorn distance function with optimal transport to learn a fair yet effective feature embedding of the nodes in the source graph. We experimentally demonstrate the effectiveness of FOCI at addressing the CNNC task while simultaneously mitigating unfairness compared to other baseline methods.
We develop a general and practical framework to address the problem of the optimal design of dynamic fee mechanisms for multiple blockchain resources. Our framework allows to compute policies that optimally trade-off ...
ISBN:
(纸本)9783031786754;9783031786761
We develop a general and practical framework to address the problem of the optimal design of dynamic fee mechanisms for multiple blockchain resources. Our framework allows to compute policies that optimally trade-off between adjusting resource prices to handle persistent demand shifts versus being robust to local noise in the observed block demand. In the general case with more than one resource, our optimal policies correctly handle cross-effects (complementarity and substitutability) in resource demands. We also show how these cross-effects can be used to inform resource design, i.e. combining resources into bundles that have low demand-side cross-effects can yield simpler and more efficient price-update rules. Our framework is also practical, we demonstrate how it can be used to refine or inform the design of heuristic fee update rules such as EIP-1559 or EIP-4844 with two case studies. We then estimate a uni-dimensional version of our model using real market data from the Ethereum blockchain and empirically compare the performance of our optimal policies to EIP-1559.
Research on voters' trust in i-voting has been exclusively related to building trust in the process of i-voting adoption, with no work addressing the question of trust repair. This article introduces a framework f...
ISBN:
(纸本)9783031722431;9783031722448
Research on voters' trust in i-voting has been exclusively related to building trust in the process of i-voting adoption, with no work addressing the question of trust repair. This article introduces a framework for trust repair in i-voting by integrating insights from trust repair in other research areas, as well as concepts developed for and used in the e-voting literature. The article traces the process of trust repair from the different beliefs influencing voters' trust in both the human and technological dimensions of an i-voting system, through the influence of the internal and external stakeholders, to trust violations and the i-voting organisers' strategies for trust repair and the 'arsenal' of measures at their disposal. The article highlights the importance of detecting the emergence of events that may violate trust among voters, understanding the severity and dimensions of trust violation, and strategically navigating trust repair. It also outlines open questions and identifies avenues for future research.
This Late-Breaking Work explores the potential of Interactive Digital Narratives (IDN) to foster empathy and promote positive social change, particularly towards marginalized groups. Recent controversies in the gaming...
ISBN:
(纸本)9783031784491;9783031784507
This Late-Breaking Work explores the potential of Interactive Digital Narratives (IDN) to foster empathy and promote positive social change, particularly towards marginalized groups. Recent controversies in the gaming industry underscore the challenges of balancing narrative freedom with authentic representation. This project investigates whether IDN can meaningfully engage with the complexities of diversity and inclusion. Our research centers on a game created and developed by undergraduate and graduate students, featuring a Honduran immigrant facing discrimination in her new role as hiring manager. Two studies are designed to assess the game's effectiveness in fostering empathy for this character and its potential to influence players' long-term attitudes toward immigrants. The dual goals of this project are to create a compelling, empathy-driven narrative that can be utilized in educational settings and to study whether IDN can successfully increase empathy and reduce bias toward marginalized groups. This research contributes to a growing body of work on the prosocial effects of IDN and its potential to drive real-world attitude change.
Being Water is an IDN experience that speculates about the ways of being of water in the world. It combines 360 degrees video, artifact-beings, voice-over and atmospheric sound. In a first iteration, users navigate th...
ISBN:
(纸本)9783031784491;9783031784507
Being Water is an IDN experience that speculates about the ways of being of water in the world. It combines 360 degrees video, artifact-beings, voice-over and atmospheric sound. In a first iteration, users navigate through the environment to listen to authorial text. This paper's focus is on second iteration in which we experiment with integrating generative AI (genAI) to various degrees (e.g., augmenting interaction, making it replayable, sustaining a narrative) by using authorial text and scene context. We discuss preliminary findings in using LLMs as a collaborator to make meaningful additions to digital artworks.
In recent years, we have seen the growth of decentralized finance (DeFi), an ecosystem of financial applications and protocols that enable complex, automated, permissionless financial transactions in blockchains (such...
ISBN:
(纸本)9783031786754;9783031786761
In recent years, we have seen the growth of decentralized finance (DeFi), an ecosystem of financial applications and protocols that enable complex, automated, permissionless financial transactions in blockchains (such as Ethereum). We examine decentralized exchanges (DEX), a key DeFi component that facilitates token swaps. DEX prices update continuously and automatically after each swap, creating price shifts for users as their swaps (trades) wait to execute. Users protect themselves from these price shifts by setting a slippage tolerance, which represents the maximum acceptable price increase. This setting is a double-edged sword: lenient tolerance can be exploited through sandwich attacks, which cost the ecosystem over $100 million annually, but stricter tolerance may cause unnecessary failures. We perform a large-scale measurement of the impact of slippage tolerance settings on the health of the Uniswap and Sushiswap DEX ecosystems. To this end, we examine a recent change in Uniswap's default slippage setting, which aimed to mitigate sandwich attacks without increasing the likelihood of transaction failures. This change removed the prior, static default - 0.5% - in favor of one dynamically computed for each transaction based on market conditions so that sandwich attacks are less profitable. We find that, overall, Uniswap's new default slippage setting leads to a substantial reduction in Uniswap traders' losses, approximately 54.7%. The effect is even more pronounced 90% when we only consider traders who followed the default settings.
暂无评论