Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Soft Power: using cyberspace to enhance Soft Power
1. Enhancing Soft Power (ESP) Analysis and Situational Awareness over network of Computers, Sensors, People, and Content for enhancing Soft Power AmitSheth, LexisNexis Eminent Scholar Director, Ohio Center of Excellence in Knowledge-enabled Computing (Kno.e.sis) Wright State University http://knoesis.org
2. “…challenge accepted paradigms to propose new ways of fighting from air, space, and cyberspace.” - General Stephen Lorenz
3. A General who led recent war “It’s not the number of people you kill, it’s the number of people you convince.”
4. Network Analysis & Visualization Consider all challenges that can be addressed by advanced semantic analytics and visualization in the cyberspace Cyberspace = {computers/networks, sensors, social <people, conversations>} For issues effecting security in cyberspace and to enhancing military’s capacity to project soft power So broadly, People-Content-NetworkAnalysis and Visualization
5. Institutions Involved Ohio Center of Excellence in Knowledge-enabled Computing (Kno.e.sis), UCI (Ramesh Jain), OSU Faculty in Visual Analytics [Srini Parthasarthy] WSU Faculty in Internet/WWW, Semantic Web, Sensor Web, Social Computing, Network Security [Bin Wang] , Visualization [Tom Wischgoll], and Cognitive Science [John Flach,…] Optionally, faculty at Indiana University Center for Complex Networks & Systems, and Purdue University Preliminary thoughts on COE on Enhancing Soft Power (ESP)
6. Networks in Cyberspace Networks in Cyber Space Sensor Networks Computer Networks Social Networks
7. Networks and Cyber Security Networks in Cyber Space Cyber Security Sensor Networks Computer Networks Social Networks
8. Networks and Soft Power Networks in Cyber Space Computer Networks Sensor Networks Soft Power Citizen Sensing Social Networks
10. Soft Power: “The Second Face of Power” “Affect behavior through attraction” “Soft power is attractive power” Also called cooptive power: the ability to shape the preferences of others getting others to want the outcomes that we want
11. How important is soft power? Henry Crumpton– CIA covert-op leader Hard power – 25% Soft power – 75% Wield soft power (to counter terror groups in Iraq and Afghanistan) through Empowerment of native people Developmental and educational projects Measure impact of policy alternatives development initiatives
12. Using soft power and monitoring its effects One approach – Analyze online social networks Kno.e.sis Ohio State Aggregate social perceptions Analyze with a focus theme Use spatio-temporal aspects Provide sentiment, intension, cultural analysis Visualize the interaction graphs formed in social networks (more later) Twitris (live demo) How hard is this?
14. Twitris: Features and Research Issues Data intensive Processing and storage of data Vs meta data Timeliness Live resource aggregation Near real time Tunable spatio-temporal scope/weights Global Iran election Vs. Local health care debate Parallel programming, cloud computing (map-reduce) Information retrieval, natural language processing (harder with casual text), machine learning Semantic Web technologies, Web 2.0 tools
15. Twitris: Ongoing investigations For a given viral “tweet” study network of people who reply or re-tweet do this across space and theme variation of content HCR – Health care reform, IE – Iran election, ISWC – Intl. Semantic Web conference
16. Sparse and Dense ReTweet Graphs - Call for action vs. Information Sharing tweets Showing retweet (forward chains) of a viral tweet in the HealthCare Debate data “Join @MarkUdall @RitterForCO and @BennetForCO to support an up-or-down vote on the public option http://tr.im/Cm2u”
17. Sparse and Dense ReTweet Graphs - Call for action vs. Information Sharing tweets Showing retweet (forward chains) of a viral tweet in the Iran data “Iran Election Crisis: 10 Incredible YouTube Videos http://bit.ly/vPDLo” Both network and content are important in studying social dynamics (here, information diffusion)
18. Analysis of Conversation and Chatter Use language constructs to analyze sentiments and intentions
26. Paradigm: Overview, Zoom, Filter, and DoD Event detection proceeds iteratively: Events are computed from time Ti to time Ti+1 Clusters at time Ti compared with those at Ti+1 Use only nodes that are active at either time which leads to significant time savings. Views (map normalized measures -> colors) Coarsening using multilevel hierarchies of nodes Graph view, community view, event view, and node view
51. Visualize event data through space, time, and themeSET On the Web: Semantic Event Tracker
52. Trusted Active Perception Abstracting high level understanding (perception) from low level sensor (SIGINT) data from heterogeneous sensors E.g., recognizing a storm event in an area of interest at the time of interest from variety of hundred to thousands of heterogeneous sensor data Trusted Perception Cycle demo
60. Control based on message content in addition to origin (or destination)
61.
62. Way Ahead: Content Analysis to Identify Radicalization or Recruiting Collaborate with linguists, social scientists, and other computer scientists to analayze social network content (Twitter, Facebook, MySpace) to weed out radical elements. Alleged Fort Hood Shooter’s online post 6 months before the massacre “If one suicide bomber can kill 100 enemy soldiers because they were caught off guard that would be considered a strategic victory” - Maj. NidalMalikHasan
63. Way Ahead: Use and Detect misuse of Soft Power Web sites in a major world power are required to employ people who monitor and delete objectionable content; tens of thousands of others are paid to “guide” bulletin board Web exchanges in the government’s favor. Use social network analysis to sense sentiments Iranian reaction to Hilary Clinton’s statements on the election. Impact of giving (partial) control to an “ally” to operate drones in the area
There are three types of networks in cyber space as shown on the slide.
Cyber security encompassespreventing, detecting, and responding to attacks on computing infrastructure.For the purpose of this talk we are only concerned about network infrastructure.Computer networks connect systems whose contents are fairly stable.Sensor networks connect sensing devices to gather up-to-date information on battlefield activities (in this context).We will talk about these aspects a bit later.
Treating people who participate in social networks as belonging to the cyberspace community.Using their participation to elicit public opinion and trends which further guide government decisions.Soft power – next slidehttp://smallwarsjournal.com/blog/2009/01/the-soft-side-of-airpower/
Let’s talk about what is hard power first. (The next few slides are from an article by Joseph S. Nye Jr. (former Dean, Harvard school of government) in the publication “Compass: A journal of leadership” published by the center for public leadership, Harvard University)Power – influence the behaviors of others to get the outcomes one wantsWe can affect behavior of others byCarrots – inducing them with paymentsSticks – coercing them with threats
Attraction here is usually done through intangible assets – attractive personality (or friendly disposition in the case of a government), political values and policies that are seen as legitimate or having moral authority.Limitations of soft power – difficult to wield and takes a lot of time to see the effects. Also, hard to quantify the effects.But nevertheless useful – hard power alone can fail completely (Hitler, Saddam Hussain)
Transition to the next slide – This is all great, but in general how do we use soft power and measure its effects?
Monitoring soft power effects and leveraging citizen sensing through social network analysis and visualization
Twitris’ features – and the CS research we use/leverage
Twitris’ features – and the CS research we use/leverage
The ones marked in red all formally defined in Srini’s work.Interactive interrogation – ability to interactively select and zoom down to clusters, entities of interest, as well as specific dynamic interactions and events that govern evolution of networks over time.
The ones marked red in the second bullet are indices formally defined.Observe that events and behavioral measures are sometimes associated with clusters and sometimes with nodesStability – measures the tendency of a node to interact with the same nodes over a period of time.Sociability – measure of the number of a different interactions that a node participates inInfluence of node – indicates how other nodes behave when this node joins or leaves a cluster (do other leave when this node leaves/ do a lot of nodes join the cluster that this node joins)Popularity - of a cluster measures how many nodes does it attract
Various views are useful for different tasks. We can track a person’s activities in the network (node view),or zoom in on a community where a number of critical events have taken place (community view), orPick a particular event and get info on it across various clusters.DoD = details on demand