Paper: Multi-adaptive Natural Language Generation using Principal Component Regression

ACL ID W14-4422
Title Multi-adaptive Natural Language Generation using Principal Component Regression
Venue International Conference on Natural Language Generation
Session Main Conference
Year 2014
Authors

We present FeedbackGen, a system that uses a multi-adaptive approach to Natu- ral Language Generation. With the term ?multi-adaptive?, we refer to a system that is able to adapt its content to dif- ferent user groups simultaneously, in our case adapting to both lecturers and stu- dents. We present a novel approach to student feedback generation, which simul- taneously takes into account the prefer- ences of lecturers and students when de- termining the content to be conveyed in a feedback summary. In this framework, we utilise knowledge derived from rat- ings on feedback summaries by extract- ing the most relevant features using Prin- cipal Component Regression (PCR) anal- ysis. We then model a reward function that is used for training a Reinforcement Learning agent. Our results with stu- ...