Knowledge Exploitation and Population for Event Reasoning (KEPLER) Phase II

In Phase II of KEPLER, we will develop a prototype Knowledge Base (KB) expansion system that employs rich common-sense knowledge resources to perform high-quality scalable inference over a KB. The system will be able to (1) extract event parameters, event constructs, ontological relationships, definitional effects and inferences, and quantitative expectations from open text and structured resources; (2) reason over a production-scale KB using probabilistic soft logic in a modular pipelined reasoning framework; (3) incorporate advanced reasoning modules, such as geo-spatial reasoning, quantitative reasoning, and temporal reasoning; (4) make use of entity and event attributes for unlocking additional information through coreferential reasoning; and (5) apply a natural language question to a probabilistic KB of explicit and inferred facts and enable the user to visualize the evidence for the resulting answers.