Wanted to let you all know that I’ve posted a demo and a writeup of the proof-of-concept “game” I’ve been working on called “Perplexity”. The entire point of Perplexity was to see how far I could get using the latest natural language technologies (i.e. the ERG) in a text-based game. Really, it is just a demo of one way to plug the ERG into Prolog to execute the predicates in a microworld, along with a writeup that describes how I did it.
Two things about the writeup:
- Thing 1: As regular readers here know by now: I’m not a linguist. I’ve tried to use analogy, examples, small words, and, in some cases, invented concepts to describe and interpret the ERG so that others can use it like I have. I apologize in advance if I’ve written something that is really off…
- Thing 2: There are comment links at the bottom of the pages in the writeup, please give me any comments or feedback you have. I don’t want to misrepresent your work!
Two things about the demo:
- Thing 1: the way it is hosted right now has some performance issues. It actually runs ~30-40% faster locally on my macbook.
Thing 2: a bonus for users of this forum: When you run the demo and text something to Perplexity, you follow up by texting a “backdoor” command:
/showand hit enter. It will then spit out the MRS, scope-resolved tree, logic, Prolog, and response from Prolog so you can see what it did. How it all works is explained in the writeup!
I would not have been able to get as far as I did without all of your help in this forum and, of course, without you guys building such a cool technology! Thanks so much!
[BTW, I’m not stopping my work on it, just got to a good point to let you all see it.]