Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists...
Transcript of Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists...
![Page 1: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/1.jpg)
The Future of Informatics
How can Informatics be a new fundamental science?
What’s the pedigree of this at Edinburgh?
What good has this done lately?
What’s carrying us forward?
I’m talking from a personal viewpoint, although I believe these personal views should be widely held.
Dave Robertson
![Page 2: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/2.jpg)
Traditional Computer Science
scientists
observationsobservations predictionspredictionstheoriestheories
Hypothesis formationand testing
Analysis
definitionsdefinitions
computersystem
![Page 3: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/3.jpg)
scientistsspecificsystem
observationsobservations predictionspredictionstheoriestheoriesdefinitionsdefinitions
scientistsspecificsystem
observationsobservations predictionspredictionstheoriestheoriesdefinitionsdefinitions
Broader View of Informatics
scientistsuniversal
computationsystem
observationsobservations predictionspredictionstheoriestheories
Hypothesis formationand testing
Analysis
definitionsdefinitions
engineersapplicationsystem
observationsobservations predictionspredictionstheoriestheoriesdefinitionsdefinitions
![Page 4: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/4.jpg)
Example (1)
scientistsdistributedprocesses
processstates
processstates
validstatesvalid
statesprocess
propertiesprocess
properties
Hypothesis formationand testing
Analysis
ambientcalculus
ambientcalculus
proteomicsscientists
proteinstructure
structurepredictionsstructure
predictionsservice
reliabilityservice
reliabilitydata
qualitydata
qualityservice
orchestrationsservice
orchestrations
![Page 5: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/5.jpg)
Example (2)
scientistsdistributedprocesses
processstates
processstates
validstatesvalid
statesprocess
propertiesprocess
properties
Hypothesis formationand testing
Analysis
ambientcalculus
ambientcalculus
cliniciansclinical
decisionsupport
protocolbehavioursprotocol
behavioursclinicaladvice
clinicaladvice
clinicalprotocolsclinical
protocolsprotocol
enactmentprotocol
enactment
![Page 6: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/6.jpg)
Example (3)
scientistsdistributedprocesses
processstates
processstates
validstatesvalid
statesprocess
propertiesprocess
properties
Hypothesis formationand testing
Analysis
ambientcalculus
ambientcalculus
cellularbiologists
cellsignalling
cellsignalscell
signalscell
responsescell
responsescell
behaviourscell
behaviourssignallingpathways
signallingpathways
![Page 7: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/7.jpg)
Classical Period (1965-1984)Growing roots:• State• Logic
![Page 8: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/8.jpg)
Eclectic Period (1985-1999)
• Many sub-fields• Many institutes
Vigorous growth:
![Page 9: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/9.jpg)
Modern Period (2000-2007)Going large:• Internet scale• Ubiquitous
![Page 10: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/10.jpg)
Post-Modern Period (2008+)Open minimalism:• Common core• Computational scholarship• Symbiosis with other sciences
![Page 11: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/11.jpg)
Examples of Recent Impact
Transformation
Adaptation
Design
Concurrency
Parsing
Synthesis
Learning
Planning
Control
Perception
Data
Types
XML databases
Java generic types
Iterative compilation
On-demand multipath routing for ad hoc networks
Design of high-performance low-power micro-architectures
Algorithmic skeletons for parallel computation
Combinatory Categorical Grammar
Festival modular speech synthesis toolkit
Gaussian processes
Proof planning
Humanoid robotics
3D imaging
Complexity Game theory
![Page 12: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/12.jpg)
Scale
• Numbers of nodes• Numbers of people• Data at the edge• Bandwidth• Computing resource• Service complexity• Device connectivity
( )
Time
Level where centralisingbecomes impractical
Complexity factorsin the global network:
![Page 13: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/13.jpg)
Semantics of single knowledge base
Hybrid architectureSemantics across architecture
Open system of reasonersSemantics via interactions
1990 2010
An Effect of Scaleon Theory
![Page 14: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/14.jpg)
An Effect of Scale on Applications
• Automation is necessary to get most things to work.
• Lack of background knowledge for reasoners isn’t a showstopper.
• New internet environments bridge physical and virtual worlds.
• Many traditional algorithms don’t scale.
• Traditional boundednessassumptions no longer apply.
• No single sub-field matches these new environments.
Opportunities Problems
![Page 15: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/15.jpg)
Intersections
Electronicinstitutions
Servicechoreography
Distributedworkflow
Processcalculi
Contextualreasoning
Semanticquery routing
Deonticspecifications
Workflow
SOA
KRSE
MAS
P2P
FM
CoordinatingDistributed
Computation
![Page 16: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/16.jpg)
Examples of Responses to Scale
Using scale as a fulcrum Statistical machine translation; Web as a corpusPeer to peer Agent systems; Web service choreography
Guarantees in open systems Proof carrying codeProvenance of data Digital curation
Statistical complexity Random structures and algorithms
![Page 17: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/17.jpg)
Scale-driven Application Challenges• Healthcare: We know that centralised
architectures can’t scale to the demands of current healthcare. Can decentralised architectures?
• Emergency response: Every device that can communicate could potentially be used in coordinating emergency responses. How?
• Data analysis: The Large Synoptic Survey Telescope will produce 30 TB of data per night. What is the architecture of the system(s) to analyse this data set?
![Page 18: Dave Robertson - School of InformaticsDave Robertson Traditional Computer Science scientists observationsobservations theoriestheories predictionspredictions Hypothesis formation and](https://reader035.fdocuments.net/reader035/viewer/2022071408/60ff51274f3b71683070b5a2/html5/thumbnails/18.jpg)
The Future of Informatics Informatics is a fundamental science because it makes sense to study the information (and information processing characteristics) of every natural system. It is a new science because the core methods for doing this aren’t in the domain of traditional sciences.
We (including you as alumni) supplied a substantial part of the foundations for this; not only the technical stuff but the ethos that allowed the science to mature.
In the process we’ve been useful, to our fellow scientists and to society. As we have gained confidence in our own science we have been better able to engage with others.
We are now being carried forward by critical mass we’ve obtained (size does matter in this respect) and by the sheer scale of theproblems for which an informatics approach is the only solution.