Reputation: 3425
Not sure if this is a good place for the question but thought I would try since there are lots of smart people here.
I am just wondering if using a constructed (i.e. not developed naturally) and syntactically unambiguous human language like Lojban could be used to perform better at natural language understanding than, say, English, since it is a more logical language.
If anyone has explored this idea or has a better understanding of NLP I'd love some feedback.
Upvotes: 4
Views: 897
Reputation: 2616
I asked a similar question at the linguistics stack exchange.
The response I received was quite informative from user @Prash:
A few corrections: lojban, though a human language, is not a natural language; it is a conlang. AFAIK, there are no native speakers of lojban: that would require teaching lojban as one of the primary languages to a very young child.
Lojban is syntactically unambiguous, and only mostly unambiguous semantically. If there were a lojban programming language, this should not matter because one would avoid writing semantically ambiguous forms (like metaphors). This question has come up on various forums for lojban, Prolog, Haskell, etc., The consensus on those forums seems to be that it is possible, but no one has done it yet. Some people (e.g. 1, e.g. 2) have attempted to implement such a thing, but AFAIK, for very limited domains.
Upvotes: 1