Computational Policies in a Need to Share Environment
June 21, 2007
Microsoft PowerPoint - Need a reader? Get one here
As applications become more sophisticated, intelligent and function in open and dynamic en-vironments, they require greater degrees of decision making and autonomy. A long range vision is for societies of intelligent, adaptive, autonomous agents, but even today, we find the new lev-els of autonomy emerging in infrastructures like the Web, Grid computing, and pervasive com-puting environments. These systems exchange information about services offered and sought and negotiate for information sharing and use. A promising approach to securing these applications involves using rich computational policies grounded in shared semantic models to advise and constrain the information practices of intelligent autonomous systems. While rudimentary policy driven systems have been demonstrated, we identify a number of capabilities that could enhance their expressivity. None is completely new, but all are under-explored in practical systems. Some have been explored in existing research policy languages, others have been identified as desired or useful, and some mostly mentioned in passing. We must reduce these to practice and to evaluate them through experimentation.