Using OpenSim to visualize data

We were intrigued when we read a blog post in VizWorld about using OpenSim to visualize data in three dimensions. The idea is that you can walk around and interact with your data, seeing it from multiple angles.

We caught up with David Burden of Daden Limited via email. Here's what he had to say.

iSGTW: Where did you get the idea of using OpenSim as a medium for visualizing data?

Burden: Visualizing data in Second Life was almost one of the first things we did with the platform. Initially plotting "historical" data like moon landing sites on a globe of the moon, then real-time data from the likes of twitter and flickr, and then the one that really made people think was real-time tracking of aircraft landing at Los Angeles airport. Whilst we did try plotting some larger data sets (e.g. 1000 closest stars/galaxies) there were some real issues in Second Life around prim limits/costs, storage of data, data throughput, and plotting rates that meant it was not the ideal platform. Hence the move to OpenSim.

iSGTW: Are you aware of anyone else exploring this concept?

Burden: Green Phosphor is the only other company we know of doing things in this area, although I think that Intel and their ScienceSim space may be doing similar things.

iSGTW: What are the chief advantages to visualizing through OpenSim, in your opinion? The disadvantages?

Burden: The big advantages of using virtual worlds are a) that having an avatar helps you relate to the information better - in terms of spatial relationships and re-orientating to different views, all of which helps to improve understanding, identification of trends and outliers, and recall; b) that they are social shared spaces, so you can have other researchers/analysts in the same space with you, looking at exactly the same data and actually have you discussions sat in the data or by particular data points - which again helps with recall

Disadvantages are higher bandwidth and PC power (although not excessive), and some people take longer to get used to having an avatar and controlling their camera.

iSGTW: The visualization examples your blog gave were of fairly simple data sets. Have you thought about how OpenSim could be used to visualize more complex scientific data sets?

Burden: Yes, and as we mature the tools we'd like to work with more complex data. For instance we've been talking to a university about visualising DNA data, and we'd like to expand our 20 parsec local space model. People have also suggested things like visualising medical trials data. The aim is to create a generic toolset which can be applied to any data, either on its own, or in conjunction with a tool like Matlab (which we are using with OpenSim on another project). The challenge is to work out what sort of data/analysis this immersive visual analytics approach works well for, and where it doesn't, and also what type of representation work best for a given use - abstract, literal, metaphorical etc.

iSGTW: Is it something you are exploring solely so that you can offer it as a service, or are you planning on contributing any open source code to OpenSim?

Burden: The main aim is to develop a product that we can sell. However, some pieces of the work may be contributed back into OpenSim - particularly where they are generic platform enhancements rather than specific to dataviz. We've spun out software into the Open Source community before (the PIVOTE project for a virtual training authoring system), so are reasonably au fait with the pros and cons of the approach, and certainly want to give something back to the community that's created OpenSim.