Logic-Based NLU

In this project we aim to investigate the use of a Logic-based Natural language understating (NLU) system (e.g. the TRIPS natural language understanding system: http://www.cs.rochester.edu/research/trips/) with our virtual human architecture.
Some research questions we are trying to answer are:
1) Compare the authoring burden for a classic classification based NLU and the Logic based NLU.
2) What is the performance given the authoring effort?
3) Can a Logic-based NLU be combined effectively with classification-based NLU?