CeBIT Big Data 2012 - Alan Dormer, Science Leader, Government and Commercial Services, CSIROPresentation Transcript
Big Data Impact on the Economy and Society;Use Cases and PotentialAlan Dormer | Research Theme Leader1st November 2012 GOVERNMENT AND COMMERCIAL SERVICES
CSIROCommonwealth Scientific and Industrial Research Organisation • 6000 staff • Scientists are experts in their field • World class research • Impact on business, society and environmentHuman Services Delivery 2.0 | Nepal & Paris; Public Sector Efficiency Conference, Oct 2012| Page 2
Our business unitsFOOD, HEALTH ENERGY ENVIRONMENT INFORMATION & MANUFACTURING,& LIFE SCIENCE COMMUNICATIONS MATERIALS & INDUSTRIES MINERALS11 National Research Flagships 12 Research Divisions + National Research Facilities and Collections + Transformational Capability Platforms
Digital Productivity & Services Flagship Help business and government deliver new, faster, better services: • Efficiently and effectively • Doing old things in new ways • Doing new things in completely new waysHuman Services Delivery 2.0 | Nepal & Paris; Public Sector Efficiency Conference, Oct 2012| Page 4
What we doEvidence based policy and decision support – Analytics – OptimisationService delivery transformation – Services innovation – Business process and business rules optimisationCustomer centric services – Wrapping the service provider around the customer – Services personalisation
Big Data PotentialThe easier bits• When?• Where?The hard bit• Who?The really hard bit• What – Text – Voice – Image – Thoughts – Emotions
Two Distinct Foci Traditional • Index • Store • Extract/cross reference/clean • Analyse • Fine tune rules, policies, etc The Future • Real-time situational awareness • Optimise response • Review & save results • Discard
Human Services Delivery Research Alliance. . . doing old things in new ways “Working with CSIRO has changed Centrelink from a doing organisation to a thinking organisation” • 5 years (2009-2014) • $20M Catherine Rule Senior Executive • Major objective to support Service Delivery Reform
Service Delivery Reform Reduce Cost of Routine Customer Interactions More Effective Interventions Tailored information delivery Customers needs analysis Self help in on-line communities Measuring outcomes Improve communications Treatment response modelling Migrate more transactions to website More staff Reduce cost of Better available routine outcomes and making more customer lower staff effective interaction costs interventions Greater productivity through Service Delivery Reform Task reduction $1.4Bn investment Better outcomes $800M savings over 4 years
Analysing Needs to Focus InvestmentOptimising Customer Service | Alan Dormer | Page 10
Trigger Points• DHS have about 3500 calls per hour• In that hour people • lose their job • suffer a death in the family • become separated• These can be very damaging and expensive• How to identify and deal effectively with these life changing events?
Social Media Monitoring in GovernmentUnderstand how text analysis tools can facilitate social mediamonitoring for government sectorProvide overviews to discover: • General sentiment towards an issue • Emergent citizen perspectives • Event-driven user perspectives • Geographically-specific issuesImproving Services: • Identify where services can be improved • Identify citizens who are not served correctly • Protect the privacy of individuals
Visualisation of meaning in social media• Visual overview representing what is being said• Allow easy navigation to drill into content
Social Media Monitoring Relevant Search and Accidental Filter Mentions Not Centrelink
Listening to customers:Listen to customers anddecide whether to engage•Provides an overview of allsocial media.•Easy navigation (overview,drill-down).•Information to help staffmake decisions.•Archiving facility forreporting and analysis.Human Services Delivery 2.0 | Nepal & Paris; Public Sector Efficiency Conference, Oct 2012| Page 16
Improving social search • Groups discussions together. • Enables both overview and drilling down. • Captures what the talk is about.Human Services Delivery 2.0 | Nepal & Paris; Public Sector Efficiency Conference, Oct 2012| Page 18
Responding to Customers• Becoming active members of online forums.• Answering questions from individuals.• Correcting rumours and misinformation.• Clear protocols and clearance processes.• Importance of protecting customer privacy.Human Services Delivery 2.0 | Nepal & Paris; Public Sector Efficiency Conference, Oct 2012| Page 19
Customer Situational Awareness Social Media E-mails Media Overdue Usage payments Customer Call Centre Complaints Situation Transcripts Account New Closures AccountsCSIRO.
What to do with It Population Customer Situation Analysis Trends Resource allocation Strategic/tactical Characterise planning Marketing Outliers Customer serviceCSIRO.
What to do with It Personal Customer Situation Static Data Up sell Happy History Cross sell Characteristics Special Not so Offer Happy Case Unhappy Personal manage SentimentCSIRO.
UNSDI Gazetteer project Social protection: identifying vulnerable communities Making a difference to over 200M Indonesians23 |
Project driversMulti-sectoral information for Social Protection• Accurate• Uptodate• Timely• Integrated• Presented in meaningful waysSocial Protection“preventing, managing, and overcoming situations that adverselyaffect people’s well being. “- policies and programs to reduce poverty and vulnerability-reducing exposure, enhancing capacity to manage risks 1United Nations Research Institute For Social Development
Social Protection•Gazetteer framework• Infrastructure to bring data together•Traditional Data Sources• Income (240M people)• Crops, etc•Unconventional Data Sources• Phone records (2Bn calls/day)• Social media, etc•Analytics• Where are vulnerable populations?• Getting worse or better?• Exposure to natural hazards?
The problemThe information tsunamiNeed to rapidly integrate information from multiple sourcesEverything happens somewhereOne real world feature multiple:• representations• identitiers• versionsOne name multiple placesIntegration is inefficient, time consuming & expensive Discover Access, Extract Transform Load Use Use Discover Access Extract, Transform, Load Understand Time and effort Understand
Everything Happens Somewhere Geospatial information Implicitly geospatial Gazetteers (geo-codeable) UNSTATS Name GRP’08 $ IND03 NTB 8,080 IND05 NTT 4,769 BPS-ID Name GER ‘08 Tpop’10 003 Nusa Tenggara Barat 111.08 1,318,840 005 Nusa Tenggara Timur 112.09 335,805Spatial Data Infrastructure (SDI) One real world feature - Multiple representations identities, versions
Integrated index for SDI GAZETTEER Place Name(s), Feature Type, Location• Fundamental data sets for Australia Geographic Names (Gazetteer) - Geodetic network - Trig points - Property Boundaries - Plot # - Address – Physical/Allocated/Postal - Post codes and area names - Transportation - Bridges - Geographic Names (Gazetteer) - Elevation/Relief - Topographic features - Imagery - Administration Boundaries - Admin. area names - Hydrography / Bathymetry - Marine features - Hydrology – Surface Water Features - AHGF – Geofabric - POI – education, health infra - Map sheets
Example Use Case – Tsunami in Japan30 km exclusion zone:• How many people are inside?• How many schools inside?• Where are the exit roads; which ones are open?• How many schools, hospitals, aged care homes inside?• Which local governments areas?• Which rivers flow through?• Where are the nearest hospitals?Presentation title | Presenter name | Page 29
Linked data Agency X Agency Y Agency Z Education Transport StatisticsStatisticalInformation API API API Linked Data User Provide URI identifiers for use in apps UN SDI Spatial Information Gazetteer National SDI framework
The “Game Changers”Linked Data (using the Web better)Cloud Computing – massively scalable storage and processingLocally relevant global aggregated data setsCrowd-sourcing - Leveraging the power of the crowdOpen data and the ‘freemium’ access modelOpen standardsEvolutionary approach• Users – continue to use preferred gazetteer• Providers – no need to change underlying business process/systems• Multiple business models supportedBuilding an institutional infrastructure
ConclusionsBIG DATA•Opportunity for real-time situational awarenessDIFFERENCE, DIVERSITY and DISPERSION•Conventional and unconventional•Non-numeric data•Many data sets/many locations/many systemsOPPORTUNITY•Improve service delivery•Better policy
Thank youAlan DormerResearch Theme LeaderGovernment and Commercial Servicest +61 3 9545 8082e email@example.com www.csiro.au/science/servicesGOVERNMENT AND COMMERCIAL SERVICES