Haines, W., Gervasio, M., Blythe, J., Lerman, K., and Spaulding, A. A World Wider than the Web: End User Programming Across Multiple Domainsin No Code Required: Giving Users Tools to Transform the Web, Morgan Kaufmann, 2010.
This chapter presents Integrated Task Learning (ITL), an approach for learning procedures across domains using end-user programming (EUP). ITL provides a suite of complementary learning and reasoning capabilities for acquiring procedures. This includes inducing generalized procedures from observed demonstrations in instrumented applications, providing template-based procedure visualizations that are easily understandable to end users, and supporting procedure editing. To incorporate other domains, including Web domains, ITL also includes facilities for semantically mapping actions across domains. This discussion begins by describing each of these capabilities in turn and then describing the central engineering concept that ITL uses to facilitate cross-domain learning: pluggable domain models, which are independently generated type and action models over different application domains that can be combined to support cross-domain procedure learning. Finally, it briefly discusses some open questions that cross-domain EUP systems will need to address in the future.