|
||||||||||||||||
|
Ph.D. ThesesToward Energy-Aware Mobile Reasoning Agents for the Mobile Semantic Web
By Evan Patton
Over the past decade there has been an uptake of semantic technologies on mobile devices. The hardness of semantic representation languages, such as OWL 2 DL's 2NEXPTIME upper bound, coupled with device and user constraints requires means of controlling expectation with respect to time, energy, and power use. In this talk, I present a hardware-based methodology for measuring for an Android smartphone the energy and power costs associated with the task of instance realization in OWL 2 knowledge bases across a number of OWL 2 reasoners of differing complexity. These findings are used to develop knowledge base metrics and predictive models that can be used to decide whether local or remote reasoning is a more efficient use of resources based on the available hardware. This is culminated into a framework called MEAR, the Mobile Energy-Aware Reasoner framework, and I show how predictive models for an OWL 2 RL reasoner built on this framework significantly decreases runtime, energy, and power consumption in the median case.
|
||||||||||||||||
Research
