Adrian and Kingsley, (01)
AW
> the system really needs a category other than "controlled English".
> The rather radical design [of Executable English] is intended to avoid
> the brittleness/high-maintenance aspects of more conventional NL systems. (02)
The term 'controlled NL' is a moving target. It's impossible to state
necessary and sufficient conditions that distinguish everything called
a CNL from notations that are not called a CNL. For example, (03)
By Rolf Schwitter, http://www.aclweb.org/anthology/C10-2128 :
> One way to bridge the gap between a natural language and a formal
> language is the use of a controlled natural language (CNL) that can
> mediate between these languages. CNLs are engineered subsets of
> natural languages whose grammar and vocabulary have been restricted
> in a systematic way in order to reduce both ambiguity and complexity
> of full natural languages. (04)
This statement covers Aristotle's syllogisms, Weizenbaum's ELIZA
system, all or nearly all NL-like interfaces to computer systems
(including Siri and her friends), and your Executable English. (05)
Re Siri: To maximize ease of use, Siri has a very forgiving
front end that tries to interpret anything thrown at it. But
every Siri response is strictly controlled by a CNL -- with
a vocabulary that can be expanded by looking up words in a
dictionary or by using simple machine-learning techniques. (06)
AW
> When writing an app, the vocabulary is open, and so to a large
> extent is the syntax. (07)
That's true of many CNLs. The templates used to specify the EE
patterns can be recognized by a finite-state grammar. The pattern
matcher in ELIZA would be sufficient to parse them. (08)
AW
> Perhaps "uncontrolled English" ? (09)
That term sounds like 'unrestricted English', which is used for
the kind of English that everybody speaks and/or writes. (010)
JFS
>> The web page on Semantics for Interoperable Systems
>> addresses theoretical issues of logic, ontology, and methodologies.
>> The four section headings indicate the topics: (011)
KI
> ... It should simply be a case of putting the following to use:
>
> 1. Identifying Agents using HTTP URIs that resolve to Agent Profile
> Documents
> 2. Associating Identities with Identity Cards Claims (e.g., Public Key)
> and Storage Location Preferences using Relations defined in relevant
> ontologies
> 3. Protocol for Reading and Writing content over an HTTP network.
> 4. Document Content creation using RDF Statements that constrained
> by Access Controls (based on relations defined in an Ontology)
> 5. Protocol for Reading and Writing content over an HTTP network. (012)
That is a very useful approach for supporting an important kind
of interoperability. (013)
But it doesn't meet all requirements for intelligent systems
(as stated by McCarthy), for the Semantic Web (by Tim B-L), for
the 8 challenges for computable logic (by Alan Robinson), or the
"great divide" between computer science and IT (by Joseph Goguen). (014)
As I keep saying, I don't want to stop anyone from developing useful
systems while the theoreticians are debating "pie in the sky". (015)
But I wanted to lay out the full range of diverse possibilities,
both theoretical and practical. (016)
John (017)
_________________________________________________________________
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 (018)
|