[Top] [All Lists]

Re: [ontolog-forum] Discussion re reasoning about Time and State with RE

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Philip Jackson <philipcjacksonjr@xxxxxxxxxxx>
Date: Wed, 27 Aug 2014 09:21:07 -0400
Message-id: <SNT147-W9AF32BB676F7A2D8DD1FDC1DD0@xxxxxxx>
Overnight I thought of a few more things to say. So will add them, and interweave some comments for your post this morning.
1) Redefinition of "CNL" is a minor issue, at most. Confusion can be avoided, and innovation supported, if the definition is clear in context. "Sublanguage" may not be a good alternative, since it does not evoke "computable", and the Thompson's used it to discuss natural language in general, as well as their approach.
A better alternative might be to redefine the acronym, or use a different acronym:
CNL = computable natural language / computational natural language
NCL = natural computable language
ANL = artificial natural language
NAL = natural artificial language
By avoiding "controlled", these acronyms subsume an artificial language which doesn't constrain NL syntax and semantics.

> Chomsky's teacher, Zellig Harris, coined that term [sublanguage].
> He used it in a way  that's close to Wittgenstein's 'Sprachspiel' (language game).
> If you look at the various CNLs that have been developed, they are basically
> sublanguages or language games with a syntax and vocabulary tailored
> to a particular domain.
> Given that definition, I would say that a natural language is the
> totality of all sublanguages or language games that can be played
> with a given syntax and vocabulary. For any particular text, the
> challenge is to discover what game(s) is/are being played.
Understood. This is also consistent with how Thompson used the term sublanguage, in the few papers I've seen.
2) One advantage of the TalaMind approach, using a NAL that has unconstrained NL syntax and semantics, is that it leverages a point you've previously stated, which is quoted in the thesis: “Natural languages have words for all the operators of first-order logic, modal logic, and many logics that have yet to be invented.”
Since Tala includes these words, and NL syntax to support them, it facilitates AI conceptual processing for modality, causality, purposive reasoning, self-reference, conjecture, meta-reasoning, etc. These topics can be studied using Tala, without having to invent new formal logics, and without people having to understand new logic notations. Thus, the thesis is able to discuss many of these topics, and the prototype illustrates how they could be supported in a TalaMind architecture.
> PJ
> > I suggest that natural language syntax is often the best way of
> > (at least initially) representing natural language semantics.
> I'd go one step further: if any formal language could do better,
> evolution would have discovered it many millennia ago.
> That's one reason (among many) why I agree with the neuroscientist
> Antonio Damasio: "images [are] the main currency of our minds."
> See slides 37 to 47 of http://www.jfsowa.com/talks/goal2.pdf .
> That redefines the problem as one of mapping languages (natural
> and/or artificial) to and from images. That doesn't make it simpler,
> but I believe that it gets us on the right track.
I agree visualization is important, though believe that much higher-level reasoning can be supported directly using natural language processing. Perhaps this is one reason evolution developed natural language, and perhaps an initial internal mentalese equivalent to natural language. It is not clear that all thought ultimately reduces to images. (Do blind people think with images?)  So, the thesis leaves this as  a topic for future research.
3) There is no need for me to recommend future work on other alternatives, different from the TalaMind approach. People are very likely to work on alternatives anyway, following established approaches. Since I believe the TalaMind approach is the best way to achieve human-level AI (for reasons discussed in section 7.8), it's my responsibility to recommend it -- not that anyone necessarily cares what I recommend, of course.
4) Perhaps the use of Tala and the TalaMind approach may also be a good way to improve the Semantic Web. If Tala expressions were embedded in web pages, they could be more easily processed by TalaMind systems, adding more AI to the semantic web, and eventually supporting human-level AI in the web.
Thesis information (for readers new to the thread):

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (01)

<Prev in Thread] Current Thread [Next in Thread>