CSR: Small: Efficient and Scalable Systems Support for Mobile Group Formation, Inference, Recommendation and Classification

智能手机用户越来越多地使用他们的移动设备与朋友、家人和同事一起组织活动。为了提高此类移动群组内互动的丰富性，本研究项目调查了新的移动群组服务的创建和实用框架的开发，该框架将使诸如向具有不同兴趣的移动群组提供推荐之类的新功能成为可能。该项目建议开发新的方法来识别一群用户何时在移动社交网络中发生网络欺凌。此外，由于每天都有数以百万计的用户群体使用社交媒体进行交流，本项目将研究如何针对移动用户群体制定推荐和识别算法，以便快速向此类用户群体提供准确的推荐，并快速识别移动群体的网络欺凌行为。这个项目的研究和它的智力价值集中在推动理解移动群体动态的艺术状态，为移动用户群体创建准确的推荐和分类算法，并开发可扩展的基础设施，在此基础上可以部署新的移动群体应用程序。研究计划是由两个新颖且具有代表性的案例研究驱动的，即移动群体推荐和移动群体中的网络欺凌检测。该项目将生成一个由六个主要组件组成的集成系统框架，以支持移动群组应用：移动客户端应用；一套核心移动群组服务，包括投票、多播通信、评级和角色定义；移动群组信息的集合；一种移动群组推理引擎；移动群组推荐与预测引擎；以及一个移动群组分类引擎。整个社会可能会从这个项目中开发的新的移动群组服务和移动群组推荐和分类算法中获益，从而使移动群组用户能够更轻松、更丰富地互动。此外，研究社区将从开源基础设施的创建中受益，这些基础设施可以刺激对移动用户群体的进一步研究。该项目将为初高中学生制作STEM教育材料，招募本科生和未被充分代表的妇女和少数民族进行研究，将研究结果融入研究者的课堂，并加强跨学科研究。

Users of smartphones are increasingly using their mobile devices to organize activities as mobile groups of friends, family, and colleagues. In order to improve the richness of interaction within such mobile groups, this research project investigates the creation of new mobile group services and the development of a practical framework that will enable such novel capabilities as providing recommendations to mobile groups of people with diverse interests. This project proposes to develop novel ways to identify when cyberbullying by a group of users is occurring in mobile social networks. In addition, since there are millions of groups of users who use social media every day to communicate, this project will investigate how recommendation and identification algorithms for mobile groups of users can be made to scale so they can quickly provide accurate recommendations to such groups of users, and rapidly identify mobile group cyberbullying behavior. This project's research and its intellectual merit focuses on advancing the state of the art in understanding mobile group dynamics, creating accurate recommendation and classification algorithms for mobile groups of users, and developing scalable infrastructure upon which new mobile group applications can be deployed. The research plan is driven by two novel and representative case studies, namely mobile group recommendation and cyberbullying detection in mobile groups. The project will generate an integrated systems framework consisting of six major components to support mobile group applications: mobile client apps; a set of core mobile group services including voting, multicast communication, rating, and role definition; a collection of mobile group information; a mobile group inference engine; a mobile group recommendation and prediction engine; and a mobile group classification engine. Society as a whole may strongly benefit from new mobile group services and algorithms for mobile group recommendation and classification developed within this project, thereby enabling mobile groups of users to more easily and richly interact. Further, the research community will benefit by the creation of open source infrastructure that can stimulate further research on mobile groups of users. The project will produce STEM educational materials for middle/high school students, recruit undergraduates and underrepresented women and minorities for research, integrate the results of the research into classes taught by the investigators, and strengthen interdisciplinary research.