Natural Language Generation for Narrative Advancement in Serious Games

Our team is working on a new game that draws on the best elements of DISTIL. We are now developing a very sophisticated learning tool that allows the players to experience how their behaviours can shift the corporate culture of the working group that they are leading. There are three elements that are very exciting about this:

1. The game is essentially based on a psychological-rigorous personality model that underlies each of the digital characters. This means that there is no central 'script' and the game is going to act like a mirror that would accurately reflect what you show it; or as I like to call it, we are building an 'alternative reality in a box?'
   
2. As there is no script, and a potentially infinite set of outputs possible, the narrative advancement will take place exclusively through Natural Language Generation. This is an excellent test of the impressive technology that we've developed in partnership with the University of Ottawa.
   
3. Based on our earlier successes at implementing NLG driven output for the 'Business in Balance: Implementing ISO 14001' game, we have been working on an integrated authoring environment that can allow content writers to create and improve NLG templates. This is going to move this exciting technology out of DISTIL Labs and into the mainstream.
   

I can't wait until end-June when this is going to be released. As a bonus, the subject matter of the project is one that will make this world a better, happier, more livable place. I wish I could divulge more information here, but contractual restrictions require me to temper my enthusiasm, and focus on getting it done earlier, so that we can all see it sooner.

Assessment Has Passed the Litmus Test

DISTIL has made amazing strides over the past two years. We have a working demonstration that meaningful assessment can be carried out on the really rich data that is captured in digital games. I verified this with an external client recently who was completely blown away at the sophistication of the feedback that we can provide on their data. What really impressed them was that the data we used for the assessment was extracted from a simulation that was built completely without our involvement. It was their simulation, and their data, and this was the litmus test that established that our methods were generic enough to convert a large 'bucket' of meaningless clicks (which the data represented) to a very valuable suggestions on what the player should do to improve their performance.