Print

Print


Hi guys!

I finished my first article using my constructed grammar and sent it to a professor of logic. He told me my model looks unambiguous, but he can't believe the results without further proof it's really functional (I included some parse trees but had to keep the article concise). 

So he advised me to add a parser. I know some of you guys have made simple ones made based on predicate logic, but what I would need is something more like an NLP parser (like the Stanford Parser) but stripped off any probabilistic properties or whatever random natural language complications. 

It should do this for example:

[S [NP ~o] [VP [V ~as] [NP [PP [P ~'] [N ~o]] [PP [P ~'] [N ~o]]]]]

The tildes represent the root word and the grammatical endings are practically part of speech tags (-o for noun etc.). So basically the parser should prove that the following arrangement is syntactic in a minimalistic SVO model:

noun – verb – preposition – noun – preposition – noun,

and show in the parse tree that both prepositional phrases are related to the verb (and not one other; recursion takes place in verbs, adjectives and relative clauses but not in adverbials).

Also, is C++ standard in this kind of stuff, or would it be useful to make it with JavaScript?