NSF Grant to Profs. Dou and Lowd to Study Knowledge Translation and Integration for Intelligent Systems
Professors Dejing Dou and Daniel Lowd have been awarded a three-year, $495k grant by the National Science Foundation (IIS-Core programs) to study the problem of translating and integrating statistical knowledge for data mining and other intelligent systems. This research will help build next generation knowledge acquisition and data mining systems in a distributed and heterogeneous environment.
While humans are able to communicate their learned experiences to each other, it is much harder to share knowledge among intelligent computer systems such as fault detection systems, recommender systems, and others. The main aim of Dou and Lowd's project is to make it easier to reuse knowledge in different systems with different semantics and terminologies. In the future, this kind of knowledge reuse could lead to smarter and more reliable systems.
To accomplish this task, Dou and Lowd will use Semantic Web ontologies and Markov logic, a recently developed representation language based on first-order logic and probabilistic graphical models. Combining formal ontologies and Markov logic will allow researchers to describe the semantic differences and the knowledge itself in a single, unified model. Since the task of knowledge translation and integration is new, Dou and Lowd will also create new benchmarks to evaluate the effectiveness of their methods.
More information is available at the projects web site: http://aimlab.cs.uoregon.edu/SKTI/.