arXiv - CS - Computers and Society最新文献_第10页

Fake News Detection via Wisdom of Synthetic & Representative Crowds 通过 "合成智慧 "和 "代表性人群 "检测假新闻

arXiv - CS - Computers and Society Pub Date : 2024-08-06 DOI: arxiv-2408.03154

François t'Serstevens, Roberto Cerina, Giulia Piccillo

{"title":"Fake News Detection via Wisdom of Synthetic & Representative Crowds","authors":"François t'Serstevens, Roberto Cerina, Giulia Piccillo","doi":"arxiv-2408.03154","DOIUrl":"https://doi.org/arxiv-2408.03154","url":null,"abstract":"Social media companies have struggled to provide a democratically legitimate\u0000definition of \"Fake News\". Reliance on expert judgment has attracted criticism\u0000due to a general trust deficit and political polarisation. Approaches reliant\u0000on the ``wisdom of the crowds'' are a cost-effective, transparent and inclusive\u0000alternative. This paper provides a novel end-to-end methodology to detect fake\u0000news on X via \"wisdom of the synthetic & representative crowds\". We deploy an\u0000online survey on the Lucid platform to gather veracity assessments for a number\u0000of pandemic-related tweets from crowd-workers. Borrowing from the MrP\u0000literature, we train a Hierarchical Bayesian model to predict the veracity of\u0000each tweet from the perspective of different personae from the population of\u0000interest. We then weight the predicted veracity assessments according to a\u0000representative stratification frame, such that decisions about ``fake'' tweets\u0000are representative of the overall polity of interest. Based on these aggregated\u0000scores, we analyse a corpus of tweets and perform a second MrP to generate\u0000state-level estimates of the number of people who share fake news. We find\u0000small but statistically meaningful heterogeneity in fake news sharing across US\u0000states. At the individual-level: i. sharing fake news is generally rare, with\u0000an average sharing probability interval [0.07,0.14]; ii. strong evidence that\u0000Democrats share less fake news, accounting for a reduction in the sharing odds\u0000of [57.3%,3.9%] relative to the average user; iii. when Republican definitions\u0000of fake news are used, it is the latter who show a decrease in the propensity\u0000to share fake news worth [50.8%, 2.0%]; iv. some evidence that women share less\u0000fake news than men, an effect worth a [29.5%,4.9%] decrease.","PeriodicalId":501112,"journal":{"name":"arXiv - CS - Computers and Society","volume":"89 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141930924","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Reasons to Doubt the Impact of AI Risk Evaluations 怀疑人工智能风险评估影响的理由

arXiv - CS - Computers and Society Pub Date : 2024-08-05 DOI: arxiv-2408.02565

Gabriel Mukobi

引用次数: 0

An integrated view of Quantum Technology? Mapping Media, Business, and Policy Narratives 综合看待量子技术？映射媒体、商业和政策叙事

arXiv - CS - Computers and Society Pub Date : 2024-08-05 DOI: arxiv-2408.02236

Viktor Suter, Charles Ma, Gina Poehlmann, Miriam Meckel, Lea Steinacker

{"title":"An integrated view of Quantum Technology? Mapping Media, Business, and Policy Narratives","authors":"Viktor Suter, Charles Ma, Gina Poehlmann, Miriam Meckel, Lea Steinacker","doi":"arxiv-2408.02236","DOIUrl":"https://doi.org/arxiv-2408.02236","url":null,"abstract":"Narratives play a vital role in shaping public perceptions and policy on\u0000emerging technologies like quantum technology (QT). However, little is known\u0000about the construction and variation of QT narratives across societal domains.\u0000This study examines how QT is presented in business, media, and government\u0000texts using thematic narrative analysis. Our research design utilizes an\u0000extensive dataset of 36 government documents, 165 business reports, and 2,331\u0000media articles published over 20 years. We employ a computational social\u0000science approach, combining BERTopic modeling with qualitative assessment to\u0000extract themes and narratives. The findings show that public discourse on QT\u0000reflects prevailing social and political agendas, focusing on technical and\u0000commercial potential, global conflicts, national strategies, and social issues.\u0000Media articles provide the most balanced coverage, while business and\u0000government discourses often overlook societal implications. We discuss the\u0000ramifications for integrating QT into society and the need for wellinformed\u0000public discourse.","PeriodicalId":501112,"journal":{"name":"arXiv - CS - Computers and Society","volume":"46 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141930923","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Video-based Pedestrian and Vehicle Traffic Analysis During Football Games 足球比赛期间基于视频的行人和车辆交通分析

arXiv - CS - Computers and Society Pub Date : 2024-08-04 DOI: arxiv-2408.02146

Jacques P. Fleischer, Ryan Pallack, Ahan Mishra, Gustavo Riente de Andrade, Subhadipto Poddar, Emmanuel Posadas, Robert Schenck, Tania Banerjee, Anand Rangarajan, Sanjay Ranka

{"title":"Video-based Pedestrian and Vehicle Traffic Analysis During Football Games","authors":"Jacques P. Fleischer, Ryan Pallack, Ahan Mishra, Gustavo Riente de Andrade, Subhadipto Poddar, Emmanuel Posadas, Robert Schenck, Tania Banerjee, Anand Rangarajan, Sanjay Ranka","doi":"arxiv-2408.02146","DOIUrl":"https://doi.org/arxiv-2408.02146","url":null,"abstract":"This paper utilizes video analytics to study pedestrian and vehicle traffic\u0000behavior, focusing on analyzing traffic patterns during football gamedays. The\u0000University of Florida (UF) hosts six to seven home football games on Saturdays\u0000during the college football season, attracting significant pedestrian activity.\u0000Through video analytics, this study provides valuable insights into the impact\u0000of these events on traffic volumes and safety at intersections. Comparing\u0000pedestrian and vehicle activities on gamedays versus non-gamedays reveals\u0000differing patterns. For example, pedestrian volume substantially increases\u0000during gamedays, which is positively correlated with the probability of the\u0000away team winning. This correlation is likely because fans of the home team\u0000enjoy watching difficult games. Win probabilities as an early predictor of\u0000pedestrian volumes at intersections can be a tool to help traffic professionals\u0000anticipate traffic management needs. Pedestrian-to-vehicle (P2V) conflicts\u0000notably increase on gamedays, particularly a few hours before games start.\u0000Addressing this, a \"Barnes Dance\" movement phase within the intersection is\u0000recommended. Law enforcement presence during high-activity gamedays can help\u0000ensure pedestrian compliance and enhance safety. In contrast, we identified\u0000that vehicle-to-vehicle (V2V) conflicts generally do not increase on gamedays\u0000and may even decrease due to heightened driver caution.","PeriodicalId":501112,"journal":{"name":"arXiv - CS - Computers and Society","volume":"57 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141930998","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Dataset Scale and Societal Consistency Mediate Facial Impression Bias in Vision-Language AI 数据集规模和社会一致性调节视觉语言人工智能中的面部印象偏差

arXiv - CS - Computers and Society Pub Date : 2024-08-04 DOI: arxiv-2408.01959

Robert Wolfe, Aayushi Dangol, Alexis Hiniker, Bill Howe

{"title":"Dataset Scale and Societal Consistency Mediate Facial Impression Bias in Vision-Language AI","authors":"Robert Wolfe, Aayushi Dangol, Alexis Hiniker, Bill Howe","doi":"arxiv-2408.01959","DOIUrl":"https://doi.org/arxiv-2408.01959","url":null,"abstract":"Multimodal AI models capable of associating images and text hold promise for\u0000numerous domains, ranging from automated image captioning to accessibility\u0000applications for blind and low-vision users. However, uncertainty about bias\u0000has in some cases limited their adoption and availability. In the present work,\u0000we study 43 CLIP vision-language models to determine whether they learn\u0000human-like facial impression biases, and we find evidence that such biases are\u0000reflected across three distinct CLIP model families. We show for the first time\u0000that the the degree to which a bias is shared across a society predicts the\u0000degree to which it is reflected in a CLIP model. Human-like impressions of\u0000visually unobservable attributes, like trustworthiness and sexuality, emerge\u0000only in models trained on the largest dataset, indicating that a better fit to\u0000uncurated cultural data results in the reproduction of increasingly subtle\u0000social biases. Moreover, we use a hierarchical clustering approach to show that\u0000dataset size predicts the extent to which the underlying structure of facial\u0000impression bias resembles that of facial impression bias in humans. Finally, we\u0000show that Stable Diffusion models employing CLIP as a text encoder learn facial\u0000impression biases, and that these biases intersect with racial biases in Stable\u0000Diffusion XL-Turbo. While pretrained CLIP models may prove useful for\u0000scientific studies of bias, they will also require significant dataset curation\u0000when intended for use as general-purpose models in a zero-shot setting.","PeriodicalId":501112,"journal":{"name":"arXiv - CS - Computers and Society","volume":"158 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141931001","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

ML-EAT: A Multilevel Embedding Association Test for Interpretable and Transparent Social Science ML-EAT：可解释和透明社会科学的多层次嵌入关联测试

arXiv - CS - Computers and Society Pub Date : 2024-08-04 DOI: arxiv-2408.01966

Robert Wolfe, Alexis Hiniker, Bill Howe

{"title":"ML-EAT: A Multilevel Embedding Association Test for Interpretable and Transparent Social Science","authors":"Robert Wolfe, Alexis Hiniker, Bill Howe","doi":"arxiv-2408.01966","DOIUrl":"https://doi.org/arxiv-2408.01966","url":null,"abstract":"This research introduces the Multilevel Embedding Association Test (ML-EAT),\u0000a method designed for interpretable and transparent measurement of intrinsic\u0000bias in language technologies. The ML-EAT addresses issues of ambiguity and\u0000difficulty in interpreting the traditional EAT measurement by quantifying bias\u0000at three levels of increasing granularity: the differential association between\u0000two target concepts with two attribute concepts; the individual effect size of\u0000each target concept with two attribute concepts; and the association between\u0000each individual target concept and each individual attribute concept. Using the\u0000ML-EAT, this research defines a taxonomy of EAT patterns describing the nine\u0000possible outcomes of an embedding association test, each of which is associated\u0000with a unique EAT-Map, a novel four-quadrant visualization for interpreting the\u0000ML-EAT. Empirical analysis of static and diachronic word embeddings, GPT-2\u0000language models, and a CLIP language-and-image model shows that EAT patterns\u0000add otherwise unobservable information about the component biases that make up\u0000an EAT; reveal the effects of prompting in zero-shot models; and can also\u0000identify situations when cosine similarity is an ineffective metric, rendering\u0000an EAT unreliable. Our work contributes a method for rendering bias more\u0000observable and interpretable, improving the transparency of computational\u0000investigations into human minds and societies.","PeriodicalId":501112,"journal":{"name":"arXiv - CS - Computers and Society","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141931000","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

The Drama Machine: Simulating Character Development with LLM Agents 戏剧机器用 LLM 代理模拟角色发展

arXiv - CS - Computers and Society Pub Date : 2024-08-03 DOI: arxiv-2408.01725

Liam Magee, Vanicka Arora, Gus Gollings, Norma Lam-Saw

引用次数: 0

Advancing Mental Health Pre-Screening: A New Custom GPT for Psychological Distress Assessment 推进心理健康预检：用于心理压力评估的新定制 GPT

arXiv - CS - Computers and Society Pub Date : 2024-08-03 DOI: arxiv-2408.01614

Jinwen Tang, Yi Shang

引用次数: 0

Welfare, sustainability, and equity evaluation of the New York City Interborough Express using spatially heterogeneous mode choice models 利用空间异质模式选择模型对纽约市区间快车的福利、可持续性和公平性进行评估

arXiv - CS - Computers and Society Pub Date : 2024-08-02 DOI: arxiv-2408.01562

Hai Yang, Hongying Wu, Lauren Whang, Xiyuan Ren, Joseph Y. J. Chow

{"title":"Welfare, sustainability, and equity evaluation of the New York City Interborough Express using spatially heterogeneous mode choice models","authors":"Hai Yang, Hongying Wu, Lauren Whang, Xiyuan Ren, Joseph Y. J. Chow","doi":"arxiv-2408.01562","DOIUrl":"https://doi.org/arxiv-2408.01562","url":null,"abstract":"The Metropolitan Transit Authority (MTA) proposed building a new light rail\u0000route called the Interborough Express (IBX) to provide a direct, fast transit\u0000linkage between Queens and Brooklyn. An open-access synthetic citywide trip\u0000agenda dataset and a block-group-level mode choice model are used to assess the\u0000potential impact IBX could bring to New York City (NYC). IBX could save 28.1\u0000minutes to potential riders across the city. For travelers either going to or\u0000departing from areas close to IBX, the average time saving is projected to be\u000029.7 minutes. IBX is projected to have more than 254 thousand daily ridership\u0000after its completion (69% higher than reported in the official IBX proposal).\u0000Among those riders, more than 78 thousand people (30.8%) would come from\u0000low-income households while 165 thousand people (64.7%) would start or end\u0000along the IBX corridor. The addition of IBX would attract more than 50 thousand\u0000additional daily trips to transit mode, among which more than 16 thousand would\u0000be switched from using private vehicles, reducing potential greenhouse gas\u0000(GHG) emissions by 29.28 metric tons per day. IBX can also bring significant\u0000consumer surplus benefits to the communities, which are estimated to be $1.25\u0000USD per trip, or as high as $1.64 per trip made by a low-income traveler. While\u0000benefits are proportionately higher for lower-income users, the service does\u0000not appear to significantly reduce the proportion of travelers whose consumer\u0000surpluses fall below 10% of the population average (already quite low).","PeriodicalId":501112,"journal":{"name":"arXiv - CS - Computers and Society","volume":"194 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141930997","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Adaptive Recruitment Resource Allocation to Improve Cohort Representativeness in Participatory Biomedical Datasets 自适应招募资源分配，提高参与式生物医学数据集中的队列代表性

arXiv - CS - Computers and Society Pub Date : 2024-08-02 DOI: arxiv-2408.01375

Victor Borza, Andrew Estornell, Ellen Wright Clayton, Chien-Ju Ho, Russell Rothman, Yevgeniy Vorobeychik, Bradley Malin

{"title":"Adaptive Recruitment Resource Allocation to Improve Cohort Representativeness in Participatory Biomedical Datasets","authors":"Victor Borza, Andrew Estornell, Ellen Wright Clayton, Chien-Ju Ho, Russell Rothman, Yevgeniy Vorobeychik, Bradley Malin","doi":"arxiv-2408.01375","DOIUrl":"https://doi.org/arxiv-2408.01375","url":null,"abstract":"Large participatory biomedical studies, studies that recruit individuals to\u0000join a dataset, are gaining popularity and investment, especially for analysis\u0000by modern AI methods. Because they purposively recruit participants, these\u0000studies are uniquely able to address a lack of historical representation, an\u0000issue that has affected many biomedical datasets. In this work, we define\u0000representativeness as the similarity to a target population distribution of a\u0000set of attributes and our goal is to mirror the U.S. population across\u0000distributions of age, gender, race, and ethnicity. Many participatory studies\u0000recruit at several institutions, so we introduce a computational approach to\u0000adaptively allocate recruitment resources among sites to improve\u0000representativeness. In simulated recruitment of 10,000-participant cohorts from\u0000medical centers in the STAR Clinical Research Network, we show that our\u0000approach yields a more representative cohort than existing baselines. Thus, we\u0000highlight the value of computational modeling in guiding recruitment efforts.","PeriodicalId":501112,"journal":{"name":"arXiv - CS - Computers and Society","volume":"30 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141931006","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0