NLGen2 is an alternative natural language generator for use with RelEx within the OpenCog project. It takes the propositional input generated by the NLGInputView and produces sentences as its output. NLGen2 adopts the reverse parsing methodology in generating language and uses all and only those data sources used by RelEx. Additionally its architecture, algorithms, and data structures are based on Levelt's theory of generation, the Minimalist Program, and Simpler Syntax respectively. Basing NLGen2 on these theories is an attempt at achieveing linguistic plausibility.
This generator is the basis of a Master's Thesis. A short paper describing the overall system can be found at http://