Modelling4All: Learning through composing computer models.

As I read the following about efforts at MIT and Stanford to develop BioBricks, a modular way to develop synthetic DNA I saw a close similarity with the micro-behaviours of the Modelling4All Project.

What if this engineering approach is wrong? You can’t just break cells down into modular parts and then reassemble them, because part of what makes them function are complex interactions between parts that we don’t yet understand. What if we don’t know enough about DNA to program it as exactly as we would a computer? What if most researchers need specific parts nobody else wants, or ones more complicated than simple BioBrick parts? And ultimately, what should we try to build?

from http://www.stanfordalumni.org/news/magazine/2009/julaug/features/biology.html

These questions can be translated to our approach to building computer models out of modular pieces.  How important are interactions between parts? How often do modelling efforts require custom micro-behaviours?  In the current library micro-behaviours are “simple” but will many models need more complicated ones as well? What should we build?

Food for thought…

You can now connect with Modelling4All by:

Following @modelling4all on Twitter

Becoming a fan of our Facebook Page

Posting to our Discussion Board

Hope to see you there!

Today we updated the Modelling4All software at http://modelling4all.nsms.ox.ac.uk/Model/

The most important new feature is that micro-behaviour pages can have editable lists of micro-behaviours on them. This makes it much easier for non-programmers to customise existing micro-behaviours. The default library of micro-behaviours (Full-library in the Resources area) has been updated to take advantage of this. Documentation of how to author such micro-behaviours is at http://modelling4all.wikidot.com/modelling4all-programming-guide

Many small enhancements (including the ability to import a micro-behaviour by providing its URL) were also added. Many bugs were fixed, in particular, a problem where in some versions of FireFox the entire application became hidden has been fixed. Full details at http://code.google.com/p/modelling4all/source/list

Please discuss any issues at http://groups.google.com/group/modelling4all and report any bugs at http://code.google.com/p/modelling4all/issues/list

Explore, create, share, and enjoy!

We’ll be giving a talk about the project at the Oxford e-Science Centre at 2pm on 26 May. Details at http://www.oerc.ox.ac.uk/events/The-Modelling4All-Project-A-web-based-modelling.0

modelling4all.org has a new version of our web-based modelling software service. The main enhancements are:

1. Micro-behaviours can include editable text areas in code to simplify setting parameters.
2. Micro-behavioiurs with editable text areas can be copied simplifying the setting of different parameters for different agents.
3. Split screen option added so one can browse and build a model without flipping between the two work areas.
4. Many features previously provided by tabs are now accessed by the equivalent of a menu bar to improve usability.
5. The tabbed interface now removes tabs when full to avoid clutter and horizontal scrolling.

Enjoy.

And please send send bug reports and suggestions at http://code.google.com/p/modelling4all/

Ken Kahn and Howard Noble have recently submitted a paper to the SIMUTools’09 Conference. Comments are welcome. This is still a draft so please don’t cite it.

The Modelling4All Project — A web-based modelling tool embedded in Web 2.0

ABSTRACT
The Modelling4All Project is building a web-based tool for constructing, running, visualising, analysing, and sharing agent-based models. These models can be constructed by non-experts by composing pre-built modular components called micro-behaviours. We are attempting to seed and nurture a Web 2.0 community to support modelling. Models, micro-behaviours, lesson plans, tutorials, and other supporting material can be shared, discussed, reviewed, rated, and tagged.

Download the draft.

We just wrote a short plan about how we plan to get tagging, discussions, ratings, and the like by allowing others to embed the Modelling4All services into other web sites. See

http://modelling4all.wikidot.com/web-2-0-technical-design

It seems as though Google Gadgets could save some development time. The idea is to restrict the GWT/ AJAX development to creating just the core functionality outlined in the modelling4all design document:

http://modelling4all.wikidot.com/technical-design-document

namely the Explore, Construct and Experiment page tabs.

By creating gadgets that both pull and push data between the modelling4all and Google servers.

With respect to push we should be able to create gadgets that display models running in an iGoogle page (java applets at this stage although we’re aware of the limitations here e.g. speed to load and browser compatibility) for instance. The gadget would point to the modelling4all server and select a model that is open access and perhaps recent.

With respect to pull then it might be possible to use for instance a google groups (web forum tool) to provide forum functionality within the pages served by the modelling4all server. The idea being that we can capture a user’s post and send it to a google forum, then read the forum posts back from google servers and display them back on the modelling4all site. This same principle might also be true for tagging and rating services.

So first thing to build is to get a Netlogo model displaying in a Google gadget.

To investigate is whether we can pull data from existing google applications e.g. blogger/ groups to display within our AJAX pages (seems like there’s authenticated Atom feeds).

Million dollar question: how can we do account management most easily.