Tuesday, December 7, 2010
Tutorial on Crowdsourcing at ICSOC
I'm going to crowdsource Maja and my tutorial on Crowdsourcing at ICSOC.
Monday, December 6, 2010
COCOA 2010: Workday on Coordination, Collaboration and Ad-hoc processes
The COCOA 2010 event is now underway. Very good turnout, much richer than expected.
Ad-hoc and semi-structured business processes
This morning session is on ad-hoc and semi-structured business processes, with contributions from Accenture (Agata Opalach), IBM Research (Axel Martens and Dan Oppenheimer) and our own Hamid Motahari.
All of the presenters agree that the typical BPM-like description of work activities through workflow-like business processes is too rigid. Need to automate intelligently, and do not use BPM as the "dictator". Tools should adapt to what people do rather than the other way around. One more issue with the BPM-like approach is the fidelity of the model, and the problem of maintaining from going obsolete.
There is a trade-off between industrialization and flexibility, and probably we don't have a theory yet. How to get that trade-off right is a hard open problem.
Human interaction, coordination and ad-hoc activities
Second session of the morning saw contribution from Les Nelson, PARC, Daniela Busse, SAP labs, Alex Kass, Accenture and John Tang, Microsoft research.
Themes that emerged were the fact that processes are more and more cross-boundaries; the difference of synchronous vs. asynchronous collaboration (John presented an extended presence tool that is very much about synchro vs. Les' Mail2Tag system, that is mostly about asynchronous); information overload requiring collaborative filtering, and the tension between the top-down and bottom-up modes for regulating, or letting collaboration emerge.
Again, we need most help for things that are not planned in advance.
Lunch discussion feed
Few more presentations that we are using to feed our lunch discussion. Alex Birukou from University of Trento talked about collaboration in end-to-end publishing workflows. Dave Duggal of Consilience touched on innovative modeling techniques for processes, in the ideate collaboration framework. Florian Daniel, from the University of Trento talked about their wisdom-aware computing paradigm, aimed at helping with process design. Lav Varshney borrowed concepts from stochastic systems theory and adapted them to coordination.
Adaptive Case Management, Teamwork/Enterprise-wide Coordination and Collaboration
Fred Cummins, chair of the OMG task force on business management and integration gave an intro to case management and layed out some of the concepts and open challenges. Paul O'Neill from Singularity, gave the point of view on case management of the practitioner community, and drew clear differences between BPM and case management. Tara Matthews of IBM research Almaden drew attention on the importance of planning, designing and evaluating collaborations, and introduced extensions to the concept of collaboration personas as precise dsecritptions of hypotetical groups. Stan Rosenschein from Stanford presented a proposal for blending the best of both words between structured workflow and free-form collaboration.
An important point is that tools should provide assistance but not prescription, and it's important to define roles and create templates for flexible processes. This is the recipe for case management. Case management has been around for hundreds of years, but the term is picking up momentum now.
Tara noted that in CSCW, activity management is a very similar discipline to case management.
Final Discussion: Different perspectives (BPM,CSCW, NLP, etc) - Opportunities and Challenges
John Tang, Munindar Singh, Henk de Man from Cordys, Daniela Busse, Les Nelson and Christoph Bussler took part in the final discussion, representing the different communities of BPM, CSCW and adaptive case management. The discussion verted on different aspects of how much should processes been formalized versus how much flexibility should be allowed to users. It was felt that the discussion was a good vehicle for bringing the communities together even though the different point of view might be hard to reconcile. A list of challenges that we want to overcome will be the outcome of post-workshop discussions that we'll hold offline.
Ad-hoc and semi-structured business processes
This morning session is on ad-hoc and semi-structured business processes, with contributions from Accenture (Agata Opalach), IBM Research (Axel Martens and Dan Oppenheimer) and our own Hamid Motahari.
All of the presenters agree that the typical BPM-like description of work activities through workflow-like business processes is too rigid. Need to automate intelligently, and do not use BPM as the "dictator". Tools should adapt to what people do rather than the other way around. One more issue with the BPM-like approach is the fidelity of the model, and the problem of maintaining from going obsolete.
There is a trade-off between industrialization and flexibility, and probably we don't have a theory yet. How to get that trade-off right is a hard open problem.
Human interaction, coordination and ad-hoc activities
Second session of the morning saw contribution from Les Nelson, PARC, Daniela Busse, SAP labs, Alex Kass, Accenture and John Tang, Microsoft research.
Themes that emerged were the fact that processes are more and more cross-boundaries; the difference of synchronous vs. asynchronous collaboration (John presented an extended presence tool that is very much about synchro vs. Les' Mail2Tag system, that is mostly about asynchronous); information overload requiring collaborative filtering, and the tension between the top-down and bottom-up modes for regulating, or letting collaboration emerge.
Again, we need most help for things that are not planned in advance.
Lunch discussion feed
Few more presentations that we are using to feed our lunch discussion. Alex Birukou from University of Trento talked about collaboration in end-to-end publishing workflows. Dave Duggal of Consilience touched on innovative modeling techniques for processes, in the ideate collaboration framework. Florian Daniel, from the University of Trento talked about their wisdom-aware computing paradigm, aimed at helping with process design. Lav Varshney borrowed concepts from stochastic systems theory and adapted them to coordination.
Adaptive Case Management, Teamwork/Enterprise-wide Coordination and Collaboration
Fred Cummins, chair of the OMG task force on business management and integration gave an intro to case management and layed out some of the concepts and open challenges. Paul O'Neill from Singularity, gave the point of view on case management of the practitioner community, and drew clear differences between BPM and case management. Tara Matthews of IBM research Almaden drew attention on the importance of planning, designing and evaluating collaborations, and introduced extensions to the concept of collaboration personas as precise dsecritptions of hypotetical groups. Stan Rosenschein from Stanford presented a proposal for blending the best of both words between structured workflow and free-form collaboration.
An important point is that tools should provide assistance but not prescription, and it's important to define roles and create templates for flexible processes. This is the recipe for case management. Case management has been around for hundreds of years, but the term is picking up momentum now.
Tara noted that in CSCW, activity management is a very similar discipline to case management.
Final Discussion: Different perspectives (BPM,CSCW, NLP, etc) - Opportunities and Challenges
John Tang, Munindar Singh, Henk de Man from Cordys, Daniela Busse, Les Nelson and Christoph Bussler took part in the final discussion, representing the different communities of BPM, CSCW and adaptive case management. The discussion verted on different aspects of how much should processes been formalized versus how much flexibility should be allowed to users. It was felt that the discussion was a good vehicle for bringing the communities together even though the different point of view might be hard to reconcile. A list of challenges that we want to overcome will be the outcome of post-workshop discussions that we'll hold offline.
Monday, November 8, 2010
Workday on collaboration, coordination and ad-hoc processes at HP Labs, Dec 6th
I'm taking a break from posting about INFORMS (which still I feel is great - more about it later) to publicize a workday that some colleagues and I are organizing.
We call it COCOA 2010: Workday on collaboration, coordination and ad-hoc processes. We'll get together at HP Labs Palo Alto on Monday, Dec 6th - just before ICSOC begins in San Francisco, and taking advantage that quite a few of the experts from the service oriented computing side are going to be in town for that.
We are asking prospective participants to send us a position papers on topics that range from traditional CSCW and BPM related themes to groupware and adaptive case management. More details at the COCOA website that Hamid has put up.
We call it COCOA 2010: Workday on collaboration, coordination and ad-hoc processes. We'll get together at HP Labs Palo Alto on Monday, Dec 6th - just before ICSOC begins in San Francisco, and taking advantage that quite a few of the experts from the service oriented computing side are going to be in town for that.
We are asking prospective participants to send us a position papers on topics that range from traditional CSCW and BPM related themes to groupware and adaptive case management. More details at the COCOA website that Hamid has put up.
Sunday, November 7, 2010
Way to run a conference
I might reconsider this by the end of the INFORM annual meeting (actually by the time I leave on Tue), but this is a pretty interesting way to run a conference. The selection was just made based on the submission of an abstract. I have no sense for how many abstracts were accepted vs. submitted, but I've been really impressed so far with the quality of the presentations. Maybe we should leave all that lengthy peer review process for journals...
INFORMS 2010 annual meeting
I'm at the INFORMS 2010 meeting in Austin, TX. First time I attend, this is huge. There are up to 70 (!) parallel sessions at any time. It's a little overwhelming actually.
Wednesday, April 21, 2010
Notes from BDIM 2010 presentations
For those who couldn't attend BDIM 2010 in person, here are some notes on the presentations, courtesy of James Cusick. Enjoy, and comment.
BDIM Workshop
Air travel disruptions in Europe forced schedule changes on half of the presentations. Program was shuffled to allow people on site to present in the morning and remote people to present in the afternoon.
Claudio Bartolini, HP
on Wisdom of Crowds applied to scientific output quality metrics
- How to measure publications
- In some areas frequency is higher than others
- Search for community aspects, social analysis techniques
- Conducted analysis of submission patterns to understand research publication patterns
- Wisdom of crowds, 90 minutes to rank order all submissions (20 papers?)
- Comparison of traditional review and wisdom of crowds
- Reviewed by 3 reviewers in traditional mode, also used 14 crowd members
- One finding was that crowd vs traditional put weight on different papers
Thomas Setzer, Munich Unveristy and Seimes
Automated decision support for VM data center
- Energy consumption in data center is 50% of costs
- 90% in running and cooling
- More servers more energy
- Pack VMs will reduce energy costs
- First virtualize then merge to hosts
- Need to insure aggregrate capacity demand does not exceed host capacity
- Can also tune server uptime to meet demand, ie, shutdown hosts at night
- Ad hoc planning means you need low thresholds, not optimal, need to plan more carefully
- Using a matrix calculation, you can find the optimal demand usage profile for the workloads
- Based on daily workload profile you may see variation in what can be co-hosted
- Specifically, what can be co-hosted in the morning cannot be in the afternoon
- With hundreds of VM workloads you cannot migrate this load manually
- The v1/v2 model can help predicate and plan timeframes on allocation of workloads
- The graphs in the paper indicate how many servers are required per combined workload
Naresh, on financial model and IT cost center benefit enabling center
- How to get money for projects
- How to specify cost savings based on project spend
- You can consider IT the same as any other investment model
- Scale, scope, efficiencies are value propositions
- Use discount cash flow and NPV
- NPV = DCF – Initial Investment
- See paper for examples of savings models
- IT should not be looked at only as cost center, need to look at investment area with returns
Vladimar
BDIM and Patents
- BDIM papers do not reference any patents
- Academic researchers tend to ignore the patent databases
- Often reinvent the wheel
- Sample of survey is small but shows under use of patents
- Keyword searching in a particular domain can be problematic, this reduces patent references
Indgo – Automatable ITSM Processes
- Must have flexible management tools in order to handle changeable business requirements
- Islands of tools integrated but requires more flexibility
- Service oriented management platform includes all management tools and processes
- Provided meta-model for management of incidents and management processes
- Using SOA and the meta-model this can provide tools for service management
Hanan Lutfiya
Trust based decision making: middleware approach
- Subjective assessment to exhibit characteristics with the role
- Computational trust from on line systems
- Trust calculation, belief calculation, evidence gathering
- The point is to understand the evidence and how predictive it is of future trust conditions
Marco Casassa Mont
HP Labs, Bristol
- Economics of access management – providing decision support for investments
- Decision points must be made around economic investment tradeoffs
- How much security focus does there need to be
- Understand preferences on strategy, model simulations on impacts, predict outcomes
- Uses IAM model
- Models indicated impacts of up to 30% savings and outcome improvements
- This approach has potential for the future to be able to understand economic value of changes
Marco (second short presentation)
- See paper
- This is an extension of first paper, focus on explicit tradeoff decision making
Sven, HP, poster presentation
- Business relationships and outsourcing
- As more outsourcing occurs, the skills of IT needs to change to manage acquisition and delivery
- Supply chain thinking around frameworks and IT delivery services
- SCOR is one such model
- IT contract management, financials, administration are being explored by tools in the framework and supply chain of IT
- Using key points, a concept graph is created, but associated functionality to action graph is requied
- This leads to logical guidance model of activities and tasks which can be automated or have a workflow conversation defined
- This allows for some intelligence to be layered on top of Sharepoint or Wiki’s to intelligently route work from a framework perspective (ITIL and Supplier Management)
- Some things cannot be automated, ie, negotiations, procurement decisions, however, this system guides this approach and contains the workflow for the process
Gleison Baioco
Configuration management ontology
- How to relate business processes and IT configuration management artifacts
- The ontological method answers these questions
- A conceptual model is presented, an object model for configuration management
- This presentation was hard to follow due to accent problem, also, presenter read from the slides directly and there was no value add beyond that
Thoughts on BDIM future
- BDIM comes from academic background
- Need to rethink market focus on content and purpose of community
- Main companies, HP, IBM are drivers from industry
- Lack of focus on TCO
- Current theme on service quality
- For example business transformation of cloud development
Questions on IRT paper
- How can you tell if things are better
- How do you know if recurring faults are being addressed
- How do you manage change flow to prevent deployment issues
- Do you have different criteria for approving changes or is there only one process for big or small changes
- HP has done some research on mapping and visualizing ITIL tower interactions to reduce inefficiencies
- Also, demoed the tool
- Questions on why not use a vendor tool
- Also, would it not create silos of homegrown tools
BDIM Workshop
Air travel disruptions in Europe forced schedule changes on half of the presentations. Program was shuffled to allow people on site to present in the morning and remote people to present in the afternoon.
Claudio Bartolini, HP
on Wisdom of Crowds applied to scientific output quality metrics
- How to measure publications
- In some areas frequency is higher than others
- Search for community aspects, social analysis techniques
- Conducted analysis of submission patterns to understand research publication patterns
- Wisdom of crowds, 90 minutes to rank order all submissions (20 papers?)
- Comparison of traditional review and wisdom of crowds
- Reviewed by 3 reviewers in traditional mode, also used 14 crowd members
- One finding was that crowd vs traditional put weight on different papers
Thomas Setzer, Munich Unveristy and Seimes
Automated decision support for VM data center
- Energy consumption in data center is 50% of costs
- 90% in running and cooling
- More servers more energy
- Pack VMs will reduce energy costs
- First virtualize then merge to hosts
- Need to insure aggregrate capacity demand does not exceed host capacity
- Can also tune server uptime to meet demand, ie, shutdown hosts at night
- Ad hoc planning means you need low thresholds, not optimal, need to plan more carefully
- Using a matrix calculation, you can find the optimal demand usage profile for the workloads
- Based on daily workload profile you may see variation in what can be co-hosted
- Specifically, what can be co-hosted in the morning cannot be in the afternoon
- With hundreds of VM workloads you cannot migrate this load manually
- The v1/v2 model can help predicate and plan timeframes on allocation of workloads
- The graphs in the paper indicate how many servers are required per combined workload
Naresh, on financial model and IT cost center benefit enabling center
- How to get money for projects
- How to specify cost savings based on project spend
- You can consider IT the same as any other investment model
- Scale, scope, efficiencies are value propositions
- Use discount cash flow and NPV
- NPV = DCF – Initial Investment
- See paper for examples of savings models
- IT should not be looked at only as cost center, need to look at investment area with returns
Vladimar
BDIM and Patents
- BDIM papers do not reference any patents
- Academic researchers tend to ignore the patent databases
- Often reinvent the wheel
- Sample of survey is small but shows under use of patents
- Keyword searching in a particular domain can be problematic, this reduces patent references
Indgo – Automatable ITSM Processes
- Must have flexible management tools in order to handle changeable business requirements
- Islands of tools integrated but requires more flexibility
- Service oriented management platform includes all management tools and processes
- Provided meta-model for management of incidents and management processes
- Using SOA and the meta-model this can provide tools for service management
Hanan Lutfiya
Trust based decision making: middleware approach
- Subjective assessment to exhibit characteristics with the role
- Computational trust from on line systems
- Trust calculation, belief calculation, evidence gathering
- The point is to understand the evidence and how predictive it is of future trust conditions
Marco Casassa Mont
HP Labs, Bristol
- Economics of access management – providing decision support for investments
- Decision points must be made around economic investment tradeoffs
- How much security focus does there need to be
- Understand preferences on strategy, model simulations on impacts, predict outcomes
- Uses IAM model
- Models indicated impacts of up to 30% savings and outcome improvements
- This approach has potential for the future to be able to understand economic value of changes
Marco (second short presentation)
- See paper
- This is an extension of first paper, focus on explicit tradeoff decision making
Sven, HP, poster presentation
- Business relationships and outsourcing
- As more outsourcing occurs, the skills of IT needs to change to manage acquisition and delivery
- Supply chain thinking around frameworks and IT delivery services
- SCOR is one such model
- IT contract management, financials, administration are being explored by tools in the framework and supply chain of IT
- Using key points, a concept graph is created, but associated functionality to action graph is requied
- This leads to logical guidance model of activities and tasks which can be automated or have a workflow conversation defined
- This allows for some intelligence to be layered on top of Sharepoint or Wiki’s to intelligently route work from a framework perspective (ITIL and Supplier Management)
- Some things cannot be automated, ie, negotiations, procurement decisions, however, this system guides this approach and contains the workflow for the process
Gleison Baioco
Configuration management ontology
- How to relate business processes and IT configuration management artifacts
- The ontological method answers these questions
- A conceptual model is presented, an object model for configuration management
- This presentation was hard to follow due to accent problem, also, presenter read from the slides directly and there was no value add beyond that
Thoughts on BDIM future
- BDIM comes from academic background
- Need to rethink market focus on content and purpose of community
- Main companies, HP, IBM are drivers from industry
- Lack of focus on TCO
- Current theme on service quality
- For example business transformation of cloud development
Questions on IRT paper
- How can you tell if things are better
- How do you know if recurring faults are being addressed
- How do you manage change flow to prevent deployment issues
- Do you have different criteria for approving changes or is there only one process for big or small changes
- HP has done some research on mapping and visualizing ITIL tower interactions to reduce inefficiencies
- Also, demoed the tool
- Questions on why not use a vendor tool
- Also, would it not create silos of homegrown tools
BDIM 2010 Best paper award
Best paper award of BDIM 2010 goes to:Decision support for virtual machine reassignments in enterprise data centers”, Thomas Setzer (Technische Universität München, IBM T.J. Watson Research Center), Alexander Stage (Technische Universität München).
Subscribe to:
Posts (Atom)
