移動用戶行為大數(shù)據(jù)的分析及隱私保護關(guān)鍵技術(shù)研究

1、廣州市科技計劃科學(xué)研究專項申請-移動用戶行為大數(shù)據(jù)的分析及隱私保護研究一、研究目標(biāo)隨著智能移動終端和移動互聯(lián)網(wǎng)的普及應(yīng)用，日益增長的海量規(guī)模的移動用戶時空行為數(shù)據(jù)正在不斷被產(chǎn)生、傳輸和收集。這些移動數(shù)據(jù)體現(xiàn)了用戶的行為模式和日常偏好，具有重要的市場價值和社會價值。充分挖掘移動用戶行為大數(shù)據(jù)，對社會公共安全管理、市政規(guī)劃、品牌市場推廣等方面都有應(yīng)用。另一方面，移動用戶的隱私數(shù)據(jù)被對移動及人身安全嚴(yán)重負(fù)面影響，如何在充分保證用戶隱私不被的情況下，對移動大數(shù)據(jù)進(jìn)行挖掘，具有重要的科研價值和現(xiàn)實意義。本課題的研究旨在提出針對移動大數(shù)據(jù)的時間、空間、行為度特征的挖掘分析算法及，并研究移動用戶數(shù)據(jù)挖掘中的

2、隱私保護機制和算法。二、主要參與（不包括）名稱分工合作講師華南理工大學(xué)項目華南理工大學(xué)聚類算法研究趙凎森教授華南師范大學(xué)隱私安全系統(tǒng)規(guī)劃華南理工大學(xué)數(shù)據(jù)挖掘研究工程師百米生活公司數(shù)據(jù)獲取與分析三、 研究內(nèi)容1. 研究移動時空軌跡的模糊匹配檢索算法，定位滿足特定行為模式的移動用戶。該研究可用于系統(tǒng)的目標(biāo)人物追蹤應(yīng)用。目前在這方面已有一定的積累，針對真實數(shù)據(jù)進(jìn)行了分析，了文章12，并研發(fā)了一個原型系統(tǒng)，提交了一個專利申請。2. 研究移動時空數(shù)據(jù)的聚類算法，分析用戶在時間、空間、行為不同維度上的潛在關(guān)聯(lián)關(guān)系，分析移動社區(qū)的動態(tài)形成和演化機制。參與成員中的老師對集成聚類分析進(jìn)行了系統(tǒng)而富有成果的研究3

3、-8。3. 研究基于同態(tài)加密算法的移動服務(wù)隱私保護機制。研究基于云計算的高效率、低傳輸流量、輕量移動端計算的隱私保護機制。目前，針對隱私保護的通信機制，了系列文章9-12；針對分布式數(shù)據(jù)管理了系列文章14-19；針對移動服務(wù)的隱私保護，已開展了系列的前期工作，針對實際數(shù)據(jù)進(jìn)行了前期實驗，并正在撰寫文章13.四、研究基礎(chǔ)1. 數(shù)據(jù)基礎(chǔ)：移動數(shù)據(jù)：收集了億條真實的移動數(shù)據(jù)。路由器數(shù)據(jù)：百米公司未來可提供 50 萬無線路由器到的移動用戶數(shù)據(jù)樣本數(shù)據(jù)。移動 POI:20 萬廣州 POI 數(shù)據(jù)。目前已針對該數(shù)據(jù)集，設(shè)計并實現(xiàn)了隱私保護的 LBS 服務(wù)模型13.。2. 理論基礎(chǔ)：2.1 移動數(shù)據(jù)分析挖掘方

4、面：Jianming Lv, Haibiao Lin, Zhiwen Yu, Yinhong Chen, and Can, Identify and Trace1Criminals in the Crowd Aided by Fast Trajectories Retrieval, In 19thernationalConference on Database Systems for Advanced Applications (DASFAA), 2014. ( B 類會議)Haibiao Lin, Jianming Lv, Can, Miaoyi Deng, Kaitao Wang. GPS

5、 Trajectory Mining : a2Survey,Journal of Compuional Information Systems, 10(16), 2014. (EI Journal)聚類算法研究：2.2Zhiwen Yu, Le Li, Jiming Liu, Guoqiang Han, Hybrid Adaptive Classifier Ensemble, IEEE3ions on Cybernetics, 2014 (Accepted)(SCI）TranZhiwen Yu, Hongsheng Chen, Jane You, Hau-San Wong, Jiming Li

6、u, Le Li, Guoqiang Han,4Double Selection based Semi-Supervised Clustering Ensemble for Tumor Clustering fromGene ExpresProfiles, IEEE/ACM Tranions on Compuional Biology andBioinformatics, vol. 11, no. 4, pp. 1-14, 2014 (To appear). (SCI）Zhiwen Yu, Hantao Chen, Jane You, Guoqiang Han, Le Li, Hybrid F

7、uzzy Cluster Ensemble5Framework for Tumor Clustering from Bio-molecular Data, IEEE/ACM Tranions onCompuional Biology and Bioinformatics, vol. 10, no. 3, pp. 657-670, 2013.(SCI）Zhiwen Yu, Le Li, Jane You, Guoqiang Han, SC3: Triple spectral clustering based consensus6clustering framework for class dis

8、covery from cancer gene expresprofiles, IEEE/ACMTranions on Compuional Biology and Bioinformatics, vol.9, no.6, pp.1751-1765,2012. (SCI）Zhiwen Yu, Jane You, Hau-San Wong, Guoqiang Han, From Cluster Ensemble to Structure Ensemble, Information Science, vol.168, pp.81-99, 2012. (SCI 2 區(qū))8 Zhiwen Yu, Ha

9、u-San Wong, Jane You, Guoxian Yu, Guoqiang Han, “Hybrid Cluster7Ensemble Framework based on the Random Combination of Daransformation Operators”,Pattern Recognition, Vol45, Ie 5, pp.1826-1837, 2012. (SCI 2 區(qū))2.3 隱私與安全方面的相關(guān)研究：9 Jianming Lv, Tieying Zhang, Zhenhua Li, Xueqi Cheng.: Parasitic Anonymous

10、Communicationhe BitTorrent Network, under 4th round review of Computer Networks,2013.(SCI Journal, IF: 1.5)10Jianming Lv, Chaoyun Zhu, Shaohua Tang and Can. Deepflow: Hiding AnonymousCommunication Traffic in P2P Streaming Networks. Accepted by Wuhan University Journal ofNatural Scien, 2014.7.(EI Jou

11、rnal)11Jianming Lv, Jianwen Chen, Can. PACS: Parasitic Anonymous Communication inPeer-to-Peer streaming Network. Journal of Compuional Information Systems. Vol. 10 (3)2014.(EI Journal)12發(fā)明專利申請：一種寄生于對等網(wǎng)絡(luò)的通信方法。發(fā)明人：、。申請?zhí)枺?01310328952.1。13Jianming Lv. EShrink : An Encrypted Iterative Shrinkage Algorithm

12、 for Privacy2014. （寫作中）Location based Servi. Submitted to2.4 分布式數(shù)據(jù)管理相關(guān)研究：14Jianming Lv, Zhiwen Yu, Tieying Zhang, Towards An Immunity Based DistributedAlgorithm to Detect Harmful Files Shared in P2P Networks, Accepted by Peer-to-PeerNetworking and Applications. Online:10.1007/s12083-013-0221-7.(SCI

13、Journal,IF 0.4)15Jianming Lv, Can, Kaidong Liang. GRACE: A Gradient Distance Based Peer-to-PeerNetwork Supporting Efficient Content-based Retrieval.ernational Conf. on SoftComputing Techniques and Engineering Application, 2013.16Tieying Zhang, Xueqi Cheng, Jianming Lv, Zhenhua Liand Wiesong Shi, Pro

14、vidingHierarchical Lookup Service for P2P-VoD Systems, in ACM Tranions on MultimediaComputing, Communications and Applications (TOMCCAP), Vol. 8, Ie 1, No.15, Feb. 2012.(SCI Journal，IF 1.4)17Jianming Lv, Zhiwen. Yu, Tieying. Zhang. iDetect: An immunity based algorithm todetect harmful content shared

15、 in Peer-to-Peer networks. Inernational Conference onMachine Learning and Cybernetics (ICMLC), vol.2, pp.926-931, Guilin China, 2011.18Tieying Zhang, Jianming Lv, Xueqi Cheng. Mediacoop: Hierarchical Lookup for P2P-VoDServi. In 38thernational Conference on Parallel Prosing (ICPP), pp. 486-493, Vienn

16、aAustria, 2009. (B 類會議)19Jianming Lv, Xueqi Cheng. CTO: Concept Tree Based SemOverlay for PurePeer-to-Peer Information Retrieval. In 16CM Conference on Information and KnowledgeManagement (CIKM ), pp.931-934, Lisbon Portugal, 2007. (B 類會議)Jianming Lv, Xueqi Cheng, Qing Jiang and etc. LiveBT: Providing-on-demand20Streaming Service on BitTorrent Peer-to-Peer Systems. In 8th IEEEernational ConferenceTechnologies （ PDCAT ） ,on Parallel and Distributed Computing,Applicationsandpp.501-508, AdelaideAustralia, 2007.3. 項目基礎(chǔ)：申請人近年主持的主要項目：申請人近