multi-agent applications for online social network analysis€¦ · o s n more than 70% of big data...
TRANSCRIPT
Hao Lan Zhang
NIT, Zhejiang University
Multi-Agent Applications for
Online Social Network Analysis
August 2, 2013
Singapore Management University
ContentContent
Agent
Business
Applications
Online
Social
Networks
ConclusionIntroduction
& Concepts
Introduction Introduction
& Concepts& Concepts
Agen
t Conce
pts
Agen
t Conce
pts
Quoting Maes’ definition: Autonomous agents are
computational systemscomputational systemscomputational systemscomputational systemscomputational systemscomputational systemscomputational systemscomputational systems that inhabit some complex
dynamic environment, sense and act autonomously in
this environment, and by doing so realise a set of goals
or tasks for which they are designed.
Briefly: Agents are computer programscomputer programscomputer programscomputer programscomputer programscomputer programscomputer programscomputer programs that can solve
complex problems autonomously.
Agen
t Conce
pts
Agen
t Conce
pts
Quoting: Russell & Norvig, 2003, A Simple Reflex Agent
Quoting: Padgham &Winikoff2002, Agent Conceptual View
Agen
t Conce
pts
Agen
t Conce
pts
AgentsAgentsAgentsAgents’’’’ Important Features: Important Features: Important Features: Important Features: Sensors, Capabilities, Sensors, Capabilities, Sensors, Capabilities, Sensors, Capabilities, Sensors, Capabilities, Sensors, Capabilities, Sensors, Capabilities, Sensors, Capabilities,
Goals/Tasks, Reasoning & LearningGoals/Tasks, Reasoning & LearningGoals/Tasks, Reasoning & LearningGoals/Tasks, Reasoning & LearningGoals/Tasks, Reasoning & LearningGoals/Tasks, Reasoning & LearningGoals/Tasks, Reasoning & LearningGoals/Tasks, Reasoning & Learning, etc., etc., etc., etc.
AgentsAgentsAgentsAgents’’’’ Characteristics: Characteristics: Characteristics: Characteristics: Autonomous, Autonomous, Autonomous, Autonomous, Autonomous, Autonomous, Autonomous, Autonomous, Intelligent, Intelligent, Intelligent, Intelligent, Intelligent, Intelligent, Intelligent, Intelligent,
Proactive, Reactive, Mobile, Adaptive, Cooperative, Proactive, Reactive, Mobile, Adaptive, Cooperative, Proactive, Reactive, Mobile, Adaptive, Cooperative, Proactive, Reactive, Mobile, Adaptive, Cooperative, Proactive, Reactive, Mobile, Adaptive, Cooperative, Proactive, Reactive, Mobile, Adaptive, Cooperative, Proactive, Reactive, Mobile, Adaptive, Cooperative, Proactive, Reactive, Mobile, Adaptive, Cooperative,
FlexibleFlexibleFlexibleFlexibleFlexibleFlexibleFlexibleFlexible,,,, etc.etc.etc.etc.
Agent
Business
Applications
Busines
s Applic
atio
ns
Busines
s Applic
atio
ns
Agent Applications in Business:Agent Applications in Business:Agent Applications in Business:Agent Applications in Business:Agent Applications in Business:Agent Applications in Business:Agent Applications in Business:Agent Applications in Business:
●● Decision Support for Business ManagementDecision Support for Business Management
●● Online Business and Social Data AnalysisOnline Business and Social Data Analysis
●● Financial and StockFinancial and Stock--Market ApplicationsMarket Applications
●● etc.etc.
Agent Applications in Business :Agent Applications in Business :Agent Applications in Business :Agent Applications in Business :Agent Applications in Business :Agent Applications in Business :Agent Applications in Business :Agent Applications in Business :
Busines
s Applic
atio
ns
Busines
s Applic
atio
ns
JenningsJennings’’ Model of Negotiating Agents in Business Processes.Model of Negotiating Agents in Business Processes.
Busines
s Applic
atio
ns
Busines
s Applic
atio
ns
MultiMulti--agents for agents for
stock market that stock market that
can deal with stock can deal with stock
market analysis market analysis
((LuoLuo, et al)., et al).
Busines
s Applic
atio
ns
Busines
s Applic
atio
ns
MultiMulti--agents agents
for financial for financial
applications applications
that can deal that can deal
with market with market
analysis, order analysis, order
processing processing
(Ying, et al).(Ying, et al).
Busines
s Applic
atio
ns
Busines
s Applic
atio
ns
OracleOracle’’s s Enterprise Enterprise
Manager model uses Manager model uses
agents for data process agents for data process
and analysis.and analysis.
Busines
s Applic
atio
ns
Busines
s Applic
atio
ns
Agent Based Modelling for Flood Management (Newcastle UniversityAgent Based Modelling for Flood Management (Newcastle University))
Busines
s Applic
atio
ns
Busines
s Applic
atio
ns
Extended BDI agent Extended BDI agent
framework for DSS framework for DSS
(Zhang, et al).(Zhang, et al).
AgentAgent--Matrices model Matrices model
for complex problem for complex problem
solving (Zhang, et al).solving (Zhang, et al).
Online
Social
Networks
OSN
OSN
Web Sites for Online Social Networks:Web Sites for Online Social Networks:Web Sites for Online Social Networks:Web Sites for Online Social Networks:Web Sites for Online Social Networks:Web Sites for Online Social Networks:Web Sites for Online Social Networks:Web Sites for Online Social Networks:
OSN
OSN
There are 2.7 billion There are 2.7 billion There are 2.7 billion There are 2.7 billion There are 2.7 billion There are 2.7 billion There are 2.7 billion There are 2.7 billion ““““““““LikesLikesLikesLikesLikesLikesLikesLikes”””””””” every single every single every single every single every single every single every single every single
day on Facebook.day on Facebook.day on Facebook.day on Facebook.day on Facebook.day on Facebook.day on Facebook.day on Facebook.
OSN
OSN
More than 70% of More than 70% of More than 70% of More than 70% of More than 70% of More than 70% of More than 70% of More than 70% of
Big Data is generated Big Data is generated Big Data is generated Big Data is generated Big Data is generated Big Data is generated Big Data is generated Big Data is generated
by OSN and related by OSN and related by OSN and related by OSN and related by OSN and related by OSN and related by OSN and related by OSN and related
activities.activities.activities.activities.activities.activities.activities.activities.
OSN
OSN
The Evolvement of OSN:The Evolvement of OSN:The Evolvement of OSN:The Evolvement of OSN:The Evolvement of OSN:The Evolvement of OSN:The Evolvement of OSN:The Evolvement of OSN:
From: A Bunch of Locally From: A Bunch of Locally From: A Bunch of Locally From: A Bunch of Locally From: A Bunch of Locally From: A Bunch of Locally From: A Bunch of Locally From: A Bunch of Locally
Connected ComputersConnected ComputersConnected ComputersConnected ComputersConnected ComputersConnected ComputersConnected ComputersConnected ComputersTo: WorldTo: WorldTo: WorldTo: WorldTo: WorldTo: WorldTo: WorldTo: World--------Wide Connected Wide Connected Wide Connected Wide Connected Wide Connected Wide Connected Wide Connected Wide Connected
Social NetworksSocial NetworksSocial NetworksSocial NetworksSocial NetworksSocial NetworksSocial NetworksSocial Networks
OSN
OSN
Eventually, OSNs become:Eventually, OSNs become:Eventually, OSNs become:Eventually, OSNs become:Eventually, OSNs become:Eventually, OSNs become:Eventually, OSNs become:Eventually, OSNs become:
OSN
OSN
The problems emerge when online data becomes The problems emerge when online data becomes The problems emerge when online data becomes The problems emerge when online data becomes The problems emerge when online data becomes The problems emerge when online data becomes The problems emerge when online data becomes The problems emerge when online data becomes
extremely large and complex.extremely large and complex.extremely large and complex.extremely large and complex.extremely large and complex.extremely large and complex.extremely large and complex.extremely large and complex.
(1) Difficulties in processing Big(1) Difficulties in processing Big(1) Difficulties in processing Big(1) Difficulties in processing Big(1) Difficulties in processing Big(1) Difficulties in processing Big(1) Difficulties in processing Big(1) Difficulties in processing Big--------OSNOSNOSNOSNOSNOSNOSNOSN--------Data;Data;Data;Data;Data;Data;Data;Data;
(2) Increasing noise in OSN data;(2) Increasing noise in OSN data;(2) Increasing noise in OSN data;(2) Increasing noise in OSN data;(2) Increasing noise in OSN data;(2) Increasing noise in OSN data;(2) Increasing noise in OSN data;(2) Increasing noise in OSN data;
(3) Privacy and security of OSN;(3) Privacy and security of OSN;(3) Privacy and security of OSN;(3) Privacy and security of OSN;(3) Privacy and security of OSN;(3) Privacy and security of OSN;(3) Privacy and security of OSN;(3) Privacy and security of OSN;
(4) (4) (4) (4) (4) (4) (4) (4) …………………………………………
OSN
& A
NT
OSN
& A
NT
Robots ApplicationsRobots Applications Social NetworksSocial Networks World Wide WebWorld Wide Web
Similarity between Agent Network Topology Similarity between Agent Network Topology Similarity between Agent Network Topology Similarity between Agent Network Topology Similarity between Agent Network Topology Similarity between Agent Network Topology Similarity between Agent Network Topology Similarity between Agent Network Topology
(ANT) & OSN:(ANT) & OSN:(ANT) & OSN:(ANT) & OSN:(ANT) & OSN:(ANT) & OSN:(ANT) & OSN:(ANT) & OSN:
OSN
& A
NT
OSN
& A
NT
An ANT has the following characteristics:An ANT has the following characteristics:
1.1. Dynamic: ANTs are dynamic and changing forms.Dynamic: ANTs are dynamic and changing forms.
2.2. Complexity: ANTs are more complex.Complexity: ANTs are more complex.
3.3. Intelligence: Agents over an agent network has the ability to Intelligence: Agents over an agent network has the ability to
understand its ANT.understand its ANT.
4.4. Mobile: Agents over an agent network have mobility.Mobile: Agents over an agent network have mobility.
OSN
& A
NT
OSN
& A
NT
Source: Source: NatureNatureNatureNatureNatureNatureNatureNature, Watt & Strogatz, 1998, Watt & Strogatz, 1998
Complex Agent Network Topology:Complex Agent Network Topology:
OSN
& A
NT
OSN
& A
NT
Source: Source: ScienceScienceScienceScienceScienceScienceScienceScience, , BarabBarabáásisi and Albert, 1999and Albert, 1999
ScaleScaleScaleScaleScaleScaleScaleScale--------free Networksfree Networksfree Networksfree Networksfree Networksfree Networksfree Networksfree Networks
OSN
& A
NT
OSN
& A
NT
Agents can naturally deal with OSN problems since OSN Agents can naturally deal with OSN problems since OSN Agents can naturally deal with OSN problems since OSN Agents can naturally deal with OSN problems since OSN Agents can naturally deal with OSN problems since OSN Agents can naturally deal with OSN problems since OSN Agents can naturally deal with OSN problems since OSN Agents can naturally deal with OSN problems since OSN
and ANT have similaritiesand ANT have similaritiesand ANT have similaritiesand ANT have similaritiesand ANT have similaritiesand ANT have similaritiesand ANT have similaritiesand ANT have similarities........
Topological Description Language for Agent networks Topological Description Language for Agent networks Topological Description Language for Agent networks Topological Description Language for Agent networks Topological Description Language for Agent networks Topological Description Language for Agent networks Topological Description Language for Agent networks Topological Description Language for Agent networks
(TDLA) is introduced to deal with OSN problems.(TDLA) is introduced to deal with OSN problems.(TDLA) is introduced to deal with OSN problems.(TDLA) is introduced to deal with OSN problems.(TDLA) is introduced to deal with OSN problems.(TDLA) is introduced to deal with OSN problems.(TDLA) is introduced to deal with OSN problems.(TDLA) is introduced to deal with OSN problems.
OSN
& A
NT
OSN
& A
NT
Formulise simple ANT to support TDLA. Formulise simple ANT to support TDLA. Formulise simple ANT to support TDLA. Formulise simple ANT to support TDLA.
Centralised ANTs are formulised as follows:Centralised ANTs are formulised as follows:Centralised ANTs are formulised as follows:Centralised ANTs are formulised as follows:
ASPP ji −∈∃ ,
If S satisfies all the following constraints then S is a centralIf S satisfies all the following constraints then S is a centralIf S satisfies all the following constraints then S is a centralIf S satisfies all the following constraints then S is a centralised ised ised ised
ANT and A is the central agent: ANT and A is the central agent: ANT and A is the central agent: ANT and A is the central agent:
∑ =→∀ 0),(),( jiji PPfPP
1),( =∃ iPAf
1),(),(),,( −=+→∀ ∑ ∑ NPPfAPfAPP jiiji
((((PPPPiiii, , , , PPPPjjjj denote all the other denote all the other denote all the other denote all the other agents in the network except A) agents in the network except A) agents in the network except A) agents in the network except A)
SA ∈∃
Conclusion
Our
Team
Our
Team
A Joint Research Center for Social Computing A Joint Research Center for Social Computing A Joint Research Center for Social Computing A Joint Research Center for Social Computing A Joint Research Center for Social Computing A Joint Research Center for Social Computing A Joint Research Center for Social Computing A Joint Research Center for Social Computing
and Data Managementand Data Managementand Data Managementand Data Managementand Data Managementand Data Managementand Data Managementand Data Management
Our
Team
Our
Team
Our
Team
Our
Team
张彦春(Yanchun Zhang),SCDM 主任(Director),Victoria University
庞超逸(Chaoyi Pang),SCDM 执行主任(Executive Director),NIT
张浩澜(Haolan Zhang),SCDM 常务副主任(Deputy Director),NIT
汪 浩(Hao Wang),SCDM特聘研究员(Principal Researcher), NIT
于 欣,SCDM特聘研究员(Principal Researcher), NIT
李兴森,SCDM特聘研究员(Principal Researcher), NIT
何 静,SCDM特聘研究员(Principal Researcher),Victoria University
黄光燕,SCDM特聘研究员(Principal Researcher),Victoria University
疏礼兵,SCDM研究员(Researcher),NIT
沈 斌,SCDM研究员(Researcher), NIT
张浩澜,博士。职
务:总经理
[浙江大学研究生导
师,浙大宁波理工
学院副教授]
庞超逸,博士。
职务:技术研发
部副总经理
[澳大利亚联邦
科学研究院高级
研究员,河北百
人计划专家]
何静,博士。
职务:产品总
监兼技术骨干
[澳大利亚国家
优秀青年获得
者,高级研究
员]
王晓义,博士。
职务:财务副
总经理
[浙江大学宁波
理工学院财务、
审计与投资研
究所所长 ]
刘伟,博士。
职务:市场部
副总经理兼网
络营销主管
[上海理工大学
博士研究生 ]
Our
Team
Our
Team
Our
Team
Our
Team
Grants Since 2011Grants Since 2011::::::::
• 2013澳大利亚国家自然科学基金ARC (张彦春等), 探索DP项目
•• 20132013澳大利亚国家自然科学基金澳大利亚国家自然科学基金澳大利亚国家自然科学基金澳大利亚国家自然科学基金澳大利亚国家自然科学基金澳大利亚国家自然科学基金澳大利亚国家自然科学基金澳大利亚国家自然科学基金ARC (ARC (JunhuJunhu Wang, Wang, 庞超逸等庞超逸等庞超逸等庞超逸等庞超逸等庞超逸等庞超逸等庞超逸等), ), 探索探索探索探索探索探索探索探索DPDP项目项目项目项目项目项目项目项目
•• 20122012国家自然科学基金面上项目国家自然科学基金面上项目国家自然科学基金面上项目国家自然科学基金面上项目国家自然科学基金面上项目国家自然科学基金面上项目国家自然科学基金面上项目国家自然科学基金面上项目((张彦春张彦春张彦春张彦春,,,,张浩澜等张浩澜等张浩澜等张浩澜等张彦春张彦春张彦春张彦春,,,,张浩澜等张浩澜等张浩澜等张浩澜等), ), 编号编号编号编号编号编号编号编号:61272480:61272480
• 20122012宁波市自然科学基金宁波市自然科学基金宁波市自然科学基金宁波市自然科学基金宁波市自然科学基金宁波市自然科学基金宁波市自然科学基金宁波市自然科学基金((张浩澜张浩澜张浩澜张浩澜,,,,张彦春等张彦春等张彦春等张彦春等张浩澜张浩澜张浩澜张浩澜,,,,张彦春等张彦春等张彦春等张彦春等), ), 编号编号编号编号编号编号编号编号:2012A610060:2012A610060
• 2012宁波市软科学基金(张浩澜,顾德道等), 编号: 2012A10050
•• 20122012宁波市自然科学基金宁波市自然科学基金宁波市自然科学基金宁波市自然科学基金宁波市自然科学基金宁波市自然科学基金宁波市自然科学基金宁波市自然科学基金((庞超逸庞超逸庞超逸庞超逸,,,,张浩澜等张浩澜等张浩澜等张浩澜等庞超逸庞超逸庞超逸庞超逸,,,,张浩澜等张浩澜等张浩澜等张浩澜等), ), 编号编号编号编号编号编号编号编号:2012A610025:2012A610025
•• 20122012国家自然科学基金面上项目国家自然科学基金面上项目国家自然科学基金面上项目国家自然科学基金面上项目国家自然科学基金面上项目国家自然科学基金面上项目国家自然科学基金面上项目国家自然科学基金面上项目((李兴森李兴森李兴森李兴森李兴森李兴森李兴森李兴森, , 庞超逸等庞超逸等庞超逸等庞超逸等庞超逸等庞超逸等庞超逸等庞超逸等))
• 2013澳大利亚国家自然科学基金ARC (何静), 澳大利亚杰出青年项目
•• 20122012国际合作项目国际合作项目国际合作项目国际合作项目国际合作项目国际合作项目国际合作项目国际合作项目((林凯林凯林凯林凯林凯林凯林凯林凯, , 蒋大成蒋大成蒋大成蒋大成蒋大成蒋大成蒋大成蒋大成,,张浩澜等张浩澜等张浩澜等张浩澜等张浩澜等张浩澜等张浩澜等张浩澜等), ), 编号编号编号编号编号编号编号编号: 2012D10020: 2012D10020
• 2012宁波智慧城市核心引导课程项目(张浩澜等), 编号: 甬教高〔2011〕440 号
• 2011国家人社部留学人员科技活动择优项目(张浩澜), 编号:人社部 [2011]508号
• 2011浙江省哲学社会科学项目(张浩澜,李兴森等), 编号:11JCSH03YB
Our
Team
Our
Team
The members of SCDM have published extensively in top journals (The members of SCDM have published extensively in top journals (SCISCI检索检索检索检索))))检索检索检索检索))))since 2011:since 2011:
• Zhang, H.L.(张浩澜), Liu, J., Zhang, Y(张彦春). (2013) “Utilizing BDI Agents and Topological Theory for Mining on Online Social Networks”, Fundamenta Informaticae, IOS Press. (To appear, SCI & EI Indexing,ERA Core Journal, Rough Sets TOP Journal)
• Jiang, Y., Pang, C.(庞超逸), Zhang, H.L.(张浩澜), Wang, J., Li, T., Zhang, Q. and He, J(何静). (2013) “Finding the Minimum Number of Elements with Sum above a Threshold”, Information Sciences, Vol. 238, pp. 205 – 211. Elsevier Publication. (SCI & EI Indexed, Impact Factor 2.833, TOP Journal of Zhejiang University)
• Zhang, H.L. (张浩澜), Pang, C. (庞超逸) (2013) Intelligent Data Processing and Mining (Editorial), Journal of Intern. Tech., Vol.14, No.2, Taiwan Ministry of Education Publication. (SCI & EI Indexed)
• Li, X., Zhang, H.L.(张浩澜), Zhu, Z., Xiang, Z., Chen, Z., Shi, Y. (2013) “An Intelligent Transformation Knowledge Mining Method Based on Extenics”, Journal of Intern. Tech., Vol.14, No.2, Taiwan Ministry of Education Publication. (SCI & EI Indexed)
• J. He(何静), Y. Zhang(张彦春), G. Huang, P. Souza, CIRCE: Correcting Imprecise Readings and CompressingExcrescent Points for Querying Common Patterns in Uncertain Sensor Streams, Information Systems, available online 25 Jan., 2012. (SCI & EI Indexed, ERA TOP Journal)
• J. He(何静), Y. Zhang(张彦春), G. Huang, Y. Shi, J. Cao, Distributed Data Possession Checking for Securing Multiple Replicas in Geographically Dispersed Clouds, Journal of Computer and System Sciences. (SCI & EI Indexed, ERA TOP Journal)
• J. He(何静), Y. Zhang(张彦春), G. Huang, J. Cao, A Smart Web Service based on the Context of Things, ACM Transaction on Internet Technology (TOIT). 11(3): 13-23, 2012. (SCI & EI Indexed, ERA TOP Journal)
• J. He(何静), Y. Zhang(张彦春) and G. Huang, Exceptional Object Analysis for Finding Rare Environmental Events from Water Quality Datasets, Neurocomputing Journal, 92(1):69–77, 2012. (SCI & EI Indexed, ERA TOP Journal)
• Zhang, H.L.(张浩澜), Zeng, W., Van der Velden, C. (2011) “A Reusable Agent Design Pattern with Flexibility and Extensibility”, Com. Sci. & Info. Sys., Vol. 8, No. 2, pp. 1229 – 1250. ComSIS Consortium Publisher. (SCI & EI Indexed, Impact Factor 0.625)
• G. Huang, Y. Zhang(张彦春), J. He(何静), and J. Cao, Fault Tolerance in Data Gathering Wireless Sensor Networks. The Computer Journal 54(6): 976-987, 2011. (SCI & EI Indexed, ERA TOP Journal)
Our
Team
Our
Team
The Core membersThe Core members’’ publication at various conference venues in Agent Social Computpublication at various conference venues in Agent Social Computing area since 2011:ing area since 2011:
• 张浩澜, Liu, J.,Tang, Y., Pang, C. (2013) “Solving Complex Decision-Making Problems through “Agent-
Matrices” Cooperation”, WISE 2013, Springer-Verlag Press. (Accepted, ERA TOP Conference, EI Indexed)
• 张浩澜, Liu, J., Pang, C., Li, X. (2012) “An Enhanced Mechanism for Agent Capability Reuse”, LNCS, Vol.
7669, pp.115–123, Springer-Verlag Press. (EI Indexed)
• 张浩澜, Pang, C., Li, X., Shen, B. and Jiang, Y. (2012) “A Topological Description Language for Agent
Networks”, LNCS, Vol.7235, pp.759–766, Springer-Verlag Press. (EI Indexed)
• 张浩澜, Jiang, W., Wu, H., Su, L. (2012) “A Novel Automated Recognition System Based on Medical
Machining CAD Models”, LNCS, Vol.7231, pp.49–59, Springer-Verlag Press. (EI Indexed)
• Li, X., Xiang, Z., 张浩澜, Zhu, Z. (2012) “A Novel Method for Extension Transformation Knowledge
Discovery”, LNCS, Vol.7234, pp.43-50, Springer-Verlag Press. (EI Indexed)
• Li, X., Xiang, Z., 张浩澜 (2012) “An Innovation Alliance Model Based on Extenics and Multi-Agent”, WI-IAT
2012, IEEE Press. (EI Indexed)
• Jian Chen, Gansen Zhao, and张浩澜 (2012) “Social Recommendation Based on Multi-relational Analysis”, WI-
IAT 2012, IEEE Press. (EI Indexed)
• Li Mei Zhou, Yongqing Zhang, and 张浩澜 (2012) “Marketing Positioning System Designing Based on
Extenics”, WI-IAT 2012, IEEE Press. (EI Indexed)
• 张浩澜, and Christian Van der Velden (2011) “Utilizing Knowledge Based Mechanisms in Automated Feature
Recognition Processes”, LNAI, Vol. 6889, pp. 316-326, Springer-Verlag Press. (EI Indexed)
• 张浩澜, Leung, C.H.C., Tang, X. (2011) “Discovering and Matching Service Providers among Heterogeneous
Agents”, Workshop Proc. of IJCAI, pp. 27 – 32, IJCAI Workshop Publication. (IJCAI Ranked A by ERA)
Product
s
Online Social
Media Monitoring
System based on
Cloud Computing
Thank You!!!!