Crowd-io

Retire to Kerala

noreply@blogger.com (Claudio Bartolini) — Mon, 19 Aug 2013 06:32:00 +0000

I really want to retire to Kerala but not yet. The link between Kerala and this post is a little tenuous, but came to me watching this TED talk by Andrew McAfee:

Andrew McAfee is one of those speakers who can put data together that are out there for everyone to see and weave a compelling story connecting them in interesting ways. However the most interesting thread for me kind of is left hanging. It's not entirely true that (see Kerala fisheries story) when prices stabilize you can plan your future: you can only do so if you can forecast a stable income. And that's the problem that the white guy from MIT/Harvard chooses to ignore.

Presentations archive from Enterprise Search Summit

noreply@blogger.com (Claudio Bartolini) — Wed, 29 May 2013 01:42:00 +0000

Here's the presentations archive from the Enterprise Search Summit held in New York on May 21-22, 2013. Keynote by Daniel Tunkelang of Linkedin search and features a presentation from my colleagues Kas Kasravi and Marie Risov on the Enterprise Collective, a tool for finding experts in the enterprise.

Large-Scale Idea Management and Deliberation Systems worskhop

noreply@blogger.com (Claudio Bartolini) — Tue, 02 Apr 2013 03:56:00 +0000

We'll hold this year's edition of the Large-Scale Idea Management and Deliberation Systems workshop on June 29, 2013 in Munich, Germany.

Follow us at @lsdelib. Here's a brief synopsis, from the workshop page.

Business and civic organizations are seeking new technologies and services that better enable their networked communities to participate in discussions and deliberate on common issues. At the same time communities increasingly expect such opportunities to contribute to the co-design and to influence the evolution of their business and civic endeavour. Some communities need better tools for deliberating or innovating collectively; others need tools for defining shared goals and influencing their members' behaviors toward such goals (e.g., sustainability). So, while social media have greatly simplified knowledge production and sharing, a growing challenge is to develop tools for large-scale ideation and deliberation (LSID). These tools can increase societal and business effectiveness by crowdsourcing efforts and promoting participation across large communities.
This workshop aims to bring together leading researchers, designers, and engineers who are working on this new class of systems.

Facebook app privacy policy and terms of service

noreply@blogger.com (Claudio Bartolini) — Sun, 28 Oct 2012 01:01:00 +0000

I'm creating a facebook app just for fun, and to learn a little about the various aspects that are involved. A few of the things that facebook makes you take seriously are terms of services and privacy policy. And here we go. If you got redirected here by the terms of services or privacy policy of my toy app, then you accept everything as is with no guarantee that tomorrow it will still be there. I promise that I'll be careful and will not play games with your own data that my app can get access to from the facebook graph API. This is just for fun and a learning exercise.

Wil van der Aalst keynote at BPM 2012: 10 years of BPM research

noreply@blogger.com (Claudio Bartolini) — Mon, 10 Sep 2012 22:39:00 +0000

Here are some of my notes from Wil van der Aalst keynote at BPM 2012. A video recording of the talk is here.

It all started with the Petri Nets conference in the early 2000s. Keywords back then were process, workflow, activities.

Wil then proceeded to give a comprehensive classification of the work that has going on in 10 years of BPM conferences, and classified it according to 10 main scenarios (he calls them use cases) – in turn classified along the following dimensions:

Scenarios:

Obtain a model

Obtain a configurable model

Enactment

Model-only-based analysis

Log and model based analysis

Repair, extend or improve process models

Main categories on design (not surprising), enactment (not surprising), verification. Focus on this last is surprising. Wil used his classification to highlight where he thought research is needed: for example in looking at a starting model, then process data and how to improve the model. Not a lot of work in this area.

Modeling languages

Enactment infrastructure

Model analyusis

Minign (going up)

Flexibility (going down)

Reuse (going up)

Then he drew some observations which provided the community with food for thoughts:

Many papers introduce a new modeling language (needed?)
Several papers can’t be easily linked to any of the 20 scenarios
Coverage of the domain can be improved
Many papers describe implementation, but don’t make the software available
Many papers include case studies – but most use cases are sometimes artificial

(I can't help noting that all of these are applicable to most of the research communities that I've been involved with.)

Main challenges are flexibility, configuration and mining: and they are related. For example: configuration and flexibility are linked when one considers process variability at different level. Flexibility and mining are linked in the trade-off of flexibility vs conformance. Finally mining and configuration are related when one studies cross-organizational mining

To conclude, here are Wil's recommendations:

Avoid new languages without a clear purpose
Artifacts (software and data) need to be made available
Evaluate results on a pre-defined criterion, and compare with other approaches
Many prototypes are developed from scratch, and fade into oblivion – need to build on shared platforms
Contribution is not always clear

Modeling rewards and incentive mechanisms for Social BPM – Ognjen Scekic

noreply@blogger.com (Claudio Bartolini) — Tue, 04 Sep 2012 14:12:00 +0000

Ognjen works with Hong-Lihn Truong and Schahram Dustdar at TU Wien. His presentation talked about the work on "Social Computing Unit" that Schahram leads and has been working with in conjunction with Kamal Bhattacharya of IBM research (they have a good article published on the IEEE Transactions on Internet Computing).

Apart from the adaptive case management, social BPM is the other mini-megatrend (lol, kilotrend maybe?) that I'm interested in monitoring within this community. Before I get into my notes of the presentation, I want to share my feeling that this kind of research (incentive and reward setting) will take a little time for this community to digest. The main argument of it was that because we are working with free-willed agents (so-called "social computing units"), then it's important to rely not on blind obedience for work to be carried out, but on giving incentives to the participants to bring about desired behavior in the organization. However, the questions that it drew from the audience were all about how participants will game the system and how to prevent that. Kind of missing the point, I'm sorry to say.

Anyway, back to Ognjen's presentation. He started off with a couple of definitions: Social business process: business process executed by an ad-hoc assembled team of workers (crowdsourcing is a part of it). A social company is a company relying on social BPs.

Rich collaboration requires advanced crowd management techniques. This work in particular is about setting incentives (pre-work) and reward (post-work) mechanisms. In particular one needs to align the incentives of an individuals with those of an organization. An example is locationary.com.

Based on a classification of most common incentive mechanisms, Ognjen is realizing a rewarding model, which can be used as part of their social computing platform. The expressivity of the platform in encoding – and supporting execution – of rewarding mechanisms covers all the basic common leaderbording and badging mechanisms, plus common mechanisms of rewards that are used in sales for example.

They worked together with Kamal on an IT incident management system, featuring automated team compositions and monetary rewards. They also applied their social computing platform to another scenario: a rotating presidency in a board, consisting of a team of peers (on each iteration, peers vote to select among their peers the person that is most suitable to lead at that iteration).

Capabilities and level of maturity in IT-based Case Management – Jana Koehler

noreply@blogger.com (Claudio Bartolini) — Tue, 04 Sep 2012 13:36:00 +0000

This talk from Jana Koehler, Lucerne U was part of the main BPM 2012 program (process quality session). However, it fits with the various talks that were given yesterday at the adaptive case management workshop.

The main contribution of the paper was a maturity model for case management software solutions, detailed below.

Adaptive case management is a new trend in BPM . Vendors of BPM suites argue that they require: 1) more flexible guidance and 2) more context sensitive information handling.

Case management is 30 years old, rooted in social sciences and methodologies exist. Most important use cases: social work, health care and insurance. It’s inherently cross-organization (inter-institution).

Criteria:

- Complex problem
- Comprehensive coordination needs
- Joint objectives setting with client
- Interdisciplinary and inter-organizational team
- Solution individually tailored
- *** Inherently adaptive! (no need to talk about adaptive case management! Just case management tout court).
- ** Appealing metaphor for knowledge-intensive work.

Jana presented a very comprehensive case study: the case management network of Switzerland. Questions for them: what are the needs of case management, and what’s the status of case management software systems (locally organized, very subject to data protection etc…)

Now, the questions they had in mind as they went about creating the maturity model were:

- What do current IT-based case management solutions provide, and how are they used in an organization
- What is the impact of these capabilities on an organization’s practice?

Jana reviewed a 4-quadrant picture adapted from Pascal et. al. work.: knowledge complexity: high vs low; relationship complexity (high vs low). Resulting in case management, intelligent problem solving, process-oriented work; social collaboration. BPM usually works on process-oriented work. Case management is at the intersection of the high complexity of knowledge and relationship. (picture reproduced from the proceedings of BPM 2012 - courtesy of Springer)

The phases of case management are: intake -> assessment -> planning -> linking&monitoring -> evaluation

Alignment of networks as a major goal. Key characteristic: a) holistic view on complex situations b) setting objectives jointly with the client c) complex coordination, controlling and monitoring

Information handling: visualization; access (individual, role based, inter org); assessment (guided, unified, standardized).

Case history: insights (descriptive, diagnostic, predictive)

Decision making: individually taken, systematically recorded, best practices.

Collaboration and administration – integrated with emails and other prcesses, this is already solved.

From those, they spun a very simple yet comprehensive maturity model, represented in figure, again taken from the BPM proceedings. The five levels are individualistic, supported, mnaged, standardized, transformative.

The model is useful to evaluate software systems. Plus it supports organization in mastering levels of maturity exploiting benefits and addressing risks.

Current trends: adoption of mobile and cloud solution; improved collaboration (negotiation of case lead, inter-organizational sharing of sensitive data), applied business intelligence (visualizations nad cse similarity, early recognition of potential case, root case analisys), measuring and scheudlign

Major challenges: case assessment, information exchange, case similarity and case history for predictive analysis.

Considerations from Adaptive Case Management workshop at BPM 2012

noreply@blogger.com (Claudio Bartolini) — Tue, 04 Sep 2012 09:38:00 +0000

A couple of considerations after attending the workshop on adaptive case management at BPM 2012 (and unfortunately missing the discussion on why BPM falls short of being an adequate paradigm for change management).

First consideration: it all has to be put in context. We are witnessing the realization by (some in) the business process management community that flexibility and adaptation are two Very Important Aspects to be taken into account, and that case management does do a good job at that. (In particular there was a telling comment by Jana Kohler of Luzern U, who said case management is inherently adaptive. If you talk about adaptive case management, then you should review the 30 yrs literature in case management.)

Having said that, then the problems that adaptive case management deals with when contrasted with BPM are rigidity of processes (software telling people what to do rather than the other way around) and information falling through the cracks in hand-offs. And "adaptive" case management does indeed sound promising as an approach.

What I didn't see discussed - and is a definite interest of mine - was support for group deliberation within cases. That tells me that it could be a fertile intersection area for research.

Towards automated support for case management processes with declarative configurable specifications - Irina Rychkova

noreply@blogger.com (Claudio Bartolini) — Mon, 03 Sep 2012 09:31:00 +0000

Towards automated support for case management processes with declarative configurable specifications.

Irina Rychkova – CRI University Paris I – works with Selmin Nurcan

Objective: develop a notation for modeling case management processes

The main approach seems to move from an imperative approach to a more declarative one. Which is absolutely the right things to do to ensure adaptivity.

Irina’s worked example is a mortgage approval example, but she says that this approach would similarly work for other domains. (Healthcare seems the natural domain for case management. Also for police cases. The idea is different enough circumstances that change case by case.) She then uses BPMN to model all the flexibility/configurability points. The learning points is that there’s a trade-off between configurability and complexity. So she uses configurable data objects and configuration rules in BPMN.

The conclusions she draws is that declarative rules can be useful for process validation and verification, kind of like for continuous process improvement.

Towards a system support of collaborative knowledge work - Nicolas Mundbrod

noreply@blogger.com (Claudio Bartolini) — Mon, 03 Sep 2012 08:53:00 +0000

Towards a system support of collaborative knowledge work

Nicolas Mundbrod – Ulm University – works with Manfred Reichert

Why is there still no system support for knowledge work? Start from the definition of knowledge workers by Davenport, T.H., Prusak, L.: Working Knowledge. Harvard Business School Press,

Boston (2000).

Studied three use cases for collaborative knowledge workers: automotive, public authorithies and financial services. They found four objectives: deal with uncertainty, goal orientation and emergence of work (continuous planning), growing knowledge base (both individual and common).

Then they proceeded to create a model including nine dimensions along which to discriminate different types of common knowledge work, which makes for a pretty comprehensive model.

The resulting practical approach is once again based on templates – which is a reinforcement of our approach, also based on templates, which I referred to multiple times. The challenge Nicolas is tackling next is how to define these templates.

Adaptive case management from the activity modality perspective - Lars Taxen

noreply@blogger.com (Claudio Bartolini) — Mon, 03 Sep 2012 08:03:00 +0000

Lars Taxen - researcher and consultant, associated with Linkoping Uni, formerly at Ericsson

Idea paper: Adaptive Case Management from the Activity Modality Perspective.

Lars’ main argument as I understood it is that BPMN squashes the whole activity around achieving temporalization of activities, and other perspectives need to be brought back. I buy into that. In fact that paradigm results into giving too much important to prescriptive, imperative, action-oriented aspects, missing out on all the soft bits around them (info falling through the cracks in handoffs, group deliberation not integrated within the process definition, etc…)

Starts from defining Integrationism - a new theory of communication: integrating various activities of which humans are capable into creating knowledge.

BPMN 2.0 quite elaborate. Six years 140 meetings. 53 constructs of which the majority of users use about 10. Some notions remind you of the activity domain – but the main focus is on the temporalization modality – there are many other modalities that are not taken into account.

Call to action: learn how BPMN is actually used in practice. Its definition seems to have progressed without too much attention to how it’s used in practice. So that leads us to a reconceptualization of business process. People talk about human centric BPMN, data-driven BPMN, knowledge-driven BPMN.

There are other models that show other dimensions/modalities of the activity domain. For example: spatialization: where are things, how are things composed of one another? There’s ERP, PLM – product lifecycle management. All of these models try to support the integration of activities around a business objective. So we’d want to borrow from them to achieve an integrated perspective.

Towards a non-workflow theory of business processes - Ilia Bider

noreply@blogger.com (Claudio Bartolini) — Mon, 03 Sep 2012 07:30:00 +0000

Opening keynote of the adaptive case management workshop at BPM 2012 delivered by Ilia Bider, of IbisSoft and Stockholm University, Sweden.

Ilia advocated for state-based approaches as an alternative to BPMN. Where are we at with status vs. actions that need to be taken? BPMN doesn’t have a status-based representation, but an action-based one.

Define state space; define a goal (surface over the state space); define a set of permissible trajectories. Things become clearer as he moves from defining a theory to showing practical approaches that support it. The theory got mapped into a role business process support system. State space gives a shared map. Supports collaboration and coordination. Helps navigate along permissible trajectories by offering help with navigation: forcing/prohibiting/suggesting/discouraging to follow certain paths.

Then he showed it some initial implementation of test case: DealDriver. Descriptions, attachments, various history taken from a few events – the planning is mostly manual (with some automation to provide navigation through prohibition/obligations/suggestions – but admittedly not very successful in practice).

There’s another case iPB – a design studio. Shows a template approach to a sequence of steps that need to be taken around a document. Steps are not defined as tasks, they are defined as forms which define what kind of information you need to get and where to get it. The sequence is not mandatory. It’s very similar to our template-based approach defined in the IT conversation manager. The software is somehow social as it is a shared space between participants who communicate through notifications. However, it does not support linking to communication channels (skype etc...) within the system.

Adaptive case management etc etc workshop

noreply@blogger.com (Claudio Bartolini) — Mon, 03 Sep 2012 07:20:00 +0000

The program of the Adaptive case management and other non-workflow approaches to BPM is at this link: http://acm2012.blogs.dsv.su.se/. An interesting observation on the attendance: it's more diverse in gender than in age group, full of late gen-Xers by the look of it.

BPM 2012 is on in Tallinn

noreply@blogger.com (Claudio Bartolini) — Mon, 03 Sep 2012 07:08:00 +0000

Today is the first day of BPM 2012 after the PhD presentations yesterday: the workshops.
There's an embarrassment of choices - I'm attending the adaptive case management and other non-workflow approaches, though I would have liked to follow at least a couple more: process-oriented information systems and business process management and social software. (I'll sneak out in the afternoon to follow Florian Daniel's presentation at BPM and social software.)

Presentation (Prezi) on Enterprise Crowdsourcing at SummerSOC

noreply@blogger.com (Claudio Bartolini) — Fri, 27 Jul 2012 02:53:00 +0000

Here's a presentation on Enterprise Crowdsourcing that I gave at the EU Summer School on Service-Oriented Computing in Crete a few weeks ago. Some classic examples of crowdsourcing to introduce the concept to the audience, followed by a discussion on what is different about crowdsourcing in the enterprise. A lot of what I drew from is joint work with Maja Vukovic.

Spigit Icon opens to non-co-workers

noreply@blogger.com (Claudio Bartolini) — Fri, 22 Jun 2012 16:49:00 +0000

Just yesterday had I wished that Spigit Icon opened idea management campaigns to non-co-workers (that is people not having the same corporate email address domain as you). Apparently, they now do, this was announced about a week ago, but I only got notification today.

Well done, look forward to playing with it some more.

More about Icon

noreply@blogger.com (Claudio Bartolini) — Fri, 22 Jun 2012 05:03:00 +0000

I've been playing a little more with Spigit's Icon. I finally got around to understanding the idea of doing a yammer-like idea management space. That means that people I can play with are all and only people who signed up for icon with an @hp.com address.

I actually find this is the major shortcoming of an otherwise very nice platform. I was never a fan of yammer's idea of creating a social network for a company and hosting it in the cloud. I thought it may work fine for SMBs, but I never thought it might be viable for a large company (though companies like Ford appear among Yammer testimonials). In the case of icon - I would have preferred something a little more open, which I could use to invite any participant to play the games I'd set up.

When Gamification met Idea Management

noreply@blogger.com (Claudio Bartolini) — Sat, 02 Jun 2012 23:48:00 +0000

When launching idea campaigns it's probably easier to entice the crowd to submit original ideas than to get back and vote for or comment on others' ideas. So, how about making it a game (that is, applying gamification)?

A cool idea is to play ideas against one another and let participants vote on the one they like more. My colleagues from the Social Computing Lab of HP Labs had experimented with a crowdsourcing platform of theirs - that was optimized for comparing pictures. The coolest feature about it was a proprietary variation of the Elo algorithm for ranking items. (If you watched The Social Network, that's what the first FaceMash was based on.) The problem becomes one of UXP design: ok to quickly compare a picture versus another, but what if ideas have 100+ words. If people stop to think, then the addictive property of the game vanishes that makes one go click click click and essentially work for free for the campaigns. Spigit had a solution called FaceOff that I thought nailed that part of the design, though I don't seem to be able to find it through their website anymore.

If anyone knows of more systems like that, or in general about gamifying the feedback collection part of an idea campaign, I'd be interested in hearing them.

Thought of the day on idea management: campaigns need to be bounded in time and scope

noreply@blogger.com (Claudio Bartolini) — Thu, 31 May 2012 23:56:00 +0000

The thought of the day is that idea management campaigns need to be bounded both in time and in scope. It's a reflection on a comment that Josh Folk of ideascale made the other day, and it's absolutely true.

A slightly more controversial point is that one could split the phases of idea sourcing and voting on best ideas. There are pros and cons about that, and I still need to make my mind up. In any case, there's an internal campaign ongoing at the moment that my team is helping categorizing and clustering ideas, and it's evident that after some time it's lost some steam. Plus, the fact that people might not hear back about what happened to the idea they submitted can backfire...

lsdeliberation: HP labs presentation on Crowdsource Your Business Process Design

noreply@blogger.com (Claudio Bartolini) — Tue, 29 May 2012 15:38:00 +0000

Since I wasn't there to take note from my own presentation, here's the abstract of the position paper that Hamid and I submitted to Large Scale Deliberation and Idea Management workshop:

Enterprises need to continuously assess their business processes for opportunities for improvements and optimizations. Often, they hire external consulting firms to assist them in this task. This approach is costly, time-consuming, and does not guarantee an optimal result. This is because the results depend on the understanding of the existing business processes by the third party firm, and on their expertise and experiences. In this article, we propose a framework and methodology for opening up the job of business process re-design to a selected crowd with expertise in various aspects of process re-design including industrial domain experts, business process analysts and designers, and enable them to collaboratively design the new business process with the enterprise employees. This approach enables capturing the results from a diverse group of contributors, consolidating the results in one business process model and verifying the model for most common issues. Our approach relies on the wisdom of the crowd in every stage of the re-design process. We offer mechanisms for fostering the collaboration and cross-validation of results among various involved groups and the enterprise employees to ensure higher high quality results.

lsdeliberation: Maja Vukovic IBM research on Adaptive Crowdsourcing for Complex Decisions

noreply@blogger.com (Claudio Bartolini) — Tue, 29 May 2012 13:54:00 +0000

Complex decision use case: Developmnet proposal to town council, budget: 500+ million. How do you engage the silent majority? Idea: enable the mayor to coordinate the whole process.

All components exist, and the work is in integrating it. There's an adaptive process orchestration component that helps the flow move along. And there are platforms for submission of project proposals (BeyondDiscussion), mining opinions, and there's a crowdsourcing service, BizRay. BizRay had been used in isolation before, and it's now being included into this architecture.

Next steps: use decision theory to route tasks, and look into the quality issues (reputation, credibility, etc...)

lsdeliberation: Adam Westerski Universidad Politecnica de Madrid on Sentiment Analysis in idea management

noreply@blogger.com (Claudio Bartolini) — Tue, 29 May 2012 13:12:00 +0000

Gizmo: semantically empowered idea management.
Works through opinion mining. Gets a polarity and a rating from a comment.
Experiment through the Ubuntu brainstorm. Collected 21k ideas, 133k comments from 10k users. Simple algorithm counting rating of good and bad words (results in an f-measure of 67% - room for improvement, but this was just a trial: state of the art algorithm with knowledge of the domain get f-measure up in the 90s). The results are not conclusive but might improve with a better opinion mining algorithm, which is left for future work.

lsdeliberation: Josh Folk - Ideascale's ideation solutions for government

noreply@blogger.com (Claudio Bartolini) — Tue, 29 May 2012 12:45:00 +0000

Ideascale: founder came from QuestionPro (similar to ideamonkey).
Problem: How to capture qualitative feedback (such as comments in surveys).
How it works: Users submit ideas, others vote on the ideas, the best ideas bubble up.

Ideascale wants to be the best at ideation (idea sourcing) - in particular the engagement has been transformational in the US government.

A couple of usecases: crowdsourcing feedback on proposed policy + crowdsourcing internal innovation

Crowdsourcing feedback on proposing policy: the White House wanted input on the Federal Mobility Strategy. Ideascale shows statistics on posts, votes, etc, traffic sources, and a tag cloud. Ideascale also had a usecase with the FCC on promoting the national broadband plan. This one had 16,000 participations.

Crowdsourcing for internal innovation was a project with the US dept. of Veteran Affairs (VA). The VA split the phases of innovation competition into a first phase where employees submit ideas, then a phase where they just vote. At the end of the campaign they created an innovation committee to sift through the top 150 ideas, and select the top 35 to provide funding to and take to the next stage. Important think is traceability of the origin of the idea.

Suggestions:
1) attach a time to a campaign. Submitters need to know that there's an end
2) incentives. For instance, gamification-like incentive, but also lunch with the president may work.

Insights:
what matters is really participation.
specific civic requirements are: accessibility, security and privacy, watch out for irrelevant ideas
obstacles: fear of losing control
leveraging the organization best asset: employees

The World Cafe technique for group discussion

noreply@blogger.com (Claudio Bartolini) — Tue, 29 May 2012 09:44:00 +0000

Network of conversation around topics that matter.

Divide into groups, then move around, afterwards the whole group gathers.

We are trying this live wihtin the Large Scale Deliberation workshop.

lsdeliberation: Piero Fraternali, PoliMi on Community Management Systems for Social Deliberation and Action

noreply@blogger.com (Claudio Bartolini) — Tue, 29 May 2012 09:22:00 +0000

The background comes from a couple of EU projects: BPM4People and CUbRIK: human computation for multimedia content enhancement - engaging people for classifying videos etc.

Scenarios (not sentiment analysis here): structural bidirectional interaction. not only social data collection and mining, but also for action. And it becomes a content management system for users profiles for community workers and organizers.

Application scenarios: natural resource management, citizen innovation, service innovation. Make a census of the usage of social media to apply to predictions in these areas (compare UI work at WWW where they correlate the twitpics with tag "snow" with actual snow precipitation in a period).

Idea: Content Management System with common model of social users, data warehouse for community members areas, task deployment tool to manage crowd work, and social "Business Analysis Monitor" (social BAM) - what's happening real time in a given process and by extension within the social network. Example applications: early alterting, games with a purpose, crowsourced tasks.

Experiment of the social BAM: logo detection using a facebook community. Machine methods have poor precision and recall.