基于移动群智感知的物联网大数据挖掘与应用

Recent years have witnessed the explosively growing interests on the Internet of Things (IoT) from both the academia and the industry. It has becoming an overwhelming consensus that IoT is the leading force for the pending third information revolution, which makes it the focuses of national strategies of various countries, such as the “Smart Earth” of USA, the “u-Korea” and the “i-Japan”. Indeed, IoT has shown exciting successes in various application domains including telemedicine, smart home, smart city, etc. Nevertheless, how to leverage the big data generated by IoT for maximizing its intelligent business values remains a pretty much open issue. In particular, existing studies on big data mining for IoT are still in their initial stage, suffering from 1) models in fragment for very special application scenarios, 2) models that cannot handle heterogeneous information, and 3) the missing computing infrastructure. In light of this, we organize a multidiscipline team with researchers from areas of data mining and big data computing, IoT applications, and complex network analysis, to contribute an innovative research proposal on big data mining for mobile crowd-sensing IoT with applications. This project is based on our years of accumulation of data resources including the GPS data from city cars, the location data from smart phones, the check-in data of Weibo users, and the total tweets provided by the biggest four weibo platforms. Specifically, we will first build a big data mining framework for IoT, meeting the challenges rising from data decentralization and heterogeneity. We then focus on designing novel spatio-temporal context-aware mining methods and user behavioral modeling methods for highly heterogeneous information. The designed framework and models will then be exploited for supporting two important applications related to smart city: the city’s intelligent services and the city’s public securities. We finally take a network-view of the interrelationships among human, machines and the Internet. The statistical mechanics paradigm will be employed for studying the network coordination problem and the possibility of emerging intelligence. We expect to publish a series of high-quality research papers and apply for various invention patents during the project execution. A prototypical system will be also built for demonstrating the effectiveness of our methods in boosting the smart city practice.

作为第三次信息革命的带动者，物联网已成为学术界、工业界甚至国家战略的关注焦点，并在远程医疗、智能家居、智慧城市等领域获得了成功应用。尽管如此，如何更好地挖掘物联网大数据以释放其最让人激动的智能性，还是一个高度开放问题。特别地，目前已有的物联网大数据挖掘研究仍存在建模方法碎片化、异质信息未充分融合、计算架构不明等问题。有鉴于此，项目组联合来自数据挖掘、物联网应用、复杂网络分析等领域的学者，基于已有的浮动车GPS数据、手机定位数据、微博位置登入数据、四大微博全量数据等，围绕“基于移动群智感知的物联网大数据挖掘与应用”开展系统研究。项目将构建物联网大数据挖掘框架，突破时空情景和用户行为感知的数据挖掘方法，深入研究城市智能服务与公共安全等重要问题，并探讨人机物三网协调发展与共生智能问题。项目有望在物联网大数据挖掘理论与方法方面产生一批国际水平研究成果，并可为推进智慧城市建设提供技术支持和示范应用。

本项目面向国家新型智慧城市发展的重大战略需求，以及大数据与人工智能研究的国际前沿热点，深入研究了移动群智感知物联网大数据的时空建模方法及其智慧城市应用。项目组围绕“移动群智感知的核心方法”、“基于数据智能的城市计算”、“复杂网络视角的跨域传播”三个方向，开展了管理、信息与传播的多学科交叉创新研究，并在“异质数据建模”、“可解释预测”两个关键科学问题取得了重要的理论与技术突破，在国内外形成了较大学术影响。理论成果发表数据挖掘与人工智能领域的CCF-A类顶级论文近40篇、UTD管理顶刊论文和自然子刊论文各1篇、国内管理学顶刊论文4篇、城市计算专著1部，成果的技术厚度无论在管理还是计算机领域均非常突出，实现了与本领域国际同行并跑竞争。依托理论成果，项目组以城市公共卫生、公共治理和网络空间安全为中心，大力开展产学研与实践应用工作，在国家部委研发落地数据智能业务系统10余套，为抗击新冠肺炎疫情、排除城市危化品隐患、打击电信诈骗等民生重大迫切需求做出了重要贡献，获得了多次书面表彰、应用证明及成果评优。依托本项目合作平台，项目负责人吴俊杰在中国管理科学与工程一级学会成立了“人工智能技术与管理应用”专委会并担任主任委员，积极普及人工智能技术并探索人工智能的管理之道。本项目人才培养成果突出，项目负责人吴俊杰依托项目研究成果获国家杰青项目、国家创新群体项目、教育部青年长江、两项省部级科技一等奖、《管理世界》优秀论文奖等，项目组成员童咏昕获国家优青项目，王静远因在新冠疫情工作中的突出贡献获评北京市抗击新冠肺炎疫情先进个人并在新春团拜会中获总书记接见和慰问。本项目还培养博士毕业生13名、硕士生毕业生38名，4位博士生毕业后走上大学科研道路。本项目成果有望对即将到来的5 G、物联网与智慧城市的研究热潮起到很好的学术引领作用，并推动数据智能与城市计算应用进入一个新的阶段。