Questions on CLCE

[naoya_arakawa]

Dear ML members,

I'm currently doing a survey on semantic representation and have
some questions on CLCE.
I'd appreciate it if someone could answer them.

* Is there any working semantic parser for CLCE?
* Are the terms used in CLCE based on any ontology?
  And if so, is there any lexicon linking CLCE terms to the terms
  in the ontology?

Best,

--Naoya Arakawa
  JustSystems Corporation


P.S. I'm trying to change my e-mail address on this mailing list,
     but the subscription messages I sent didn't work.
     Could anyone help me?

---------------------------------------------------------------------
To unsubscribe, e-mail: cg-unsubscribe@...
For additional commands, e-mail: cg-help@...

[Arun Majumdar]

Dear Naoya,

We have a pre-alpha version of CLCE. It is working and being improved.

The terms used in the CLCE form a simple meta-model to enable the user
to define, with a boostrap set of words, any ontology or vocabulary. We
are using a Prolog top level grammar for the meta-model to compile, at
runtime, the user created terminology and then to synthesize a runtime
interpreter to read the terminology (of controlled terms and
definitions) to create CLCE structures for use in reasoning or export as
CGIF conformant to ISO 24707 (Common Logic ISO standard).

You can find out more about CLCE on John Sowa's web site:
www.jf*sowa*.com/*clce*/specs.htm

We are not writing the grammar as a DCG but have written a dependency
grammar for the meta-model using SQL inspired key terms such as "define"
or "select", "from" for the management of the terminological ontology
and Sowa's terms "declare", "and", "with" etc... We are supporting
scripts to manage the complexities for multiple-versioning of controlled
terminologies. I have attached an example of a scripts (which inherit
hierarchically as well as an output) for you as well some CLCE statement
for an example --- note the term clce(1) says I will entering only *one*
clce sentence - for illustration, this makes it easier if you want 10
entries or clce(*) for an infinite input loop that you can type a
*modal* directive:

?- clce(1).

CLCE input: 'A manager m1 of an employee x1 works for a boss y1.'

[ manager *m1 ] [ employee *x1 ] ( of ?m1 ?x1 ) ( expr ?m1 ?g_5 )
[ work *g_5 ] [ boss *y1 ] ( for ?g_5 ?y1 ) ( thme ?g_5 ?y1 )

-yes

We have tested the system by *importing* in Wordnet, CoreLex, Roget and
the CLSI Verb Ontology. We are removing behavioral bugs at the present
time and expect to have fairly nice version complete within several weeks.

Please send me your new email address if you are not getting the system
to work - I also work on supporting the CG sites and will fix it
directly for you.

Thanks,

Arun Majumdar,
VivoMind Intelligence Inc.

naoya_arakawa@... wrote:

> Dear ML members,
>
> I'm currently doing a survey on semantic representation and have
> some questions on CLCE.
> I'd appreciate it if someone could answer them.
>
> * Is there any working semantic parser for CLCE?
> * Are the terms used in CLCE based on any ontology?
>   And if so, is there any lexicon linking CLCE terms to the terms
>   in the ontology?
>
> Best,
>
> --Naoya Arakawa
>   JustSystems Corporation
>
>
> P.S. I'm trying to change my e-mail address on this mailing list,
>      but the subscription messages I sent didn't work.
>      Could anyone help me?
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: cg-unsubscribe@...
> For additional commands, e-mail: cg-help@...
>
>
>
>  

spool script.out
--
-- Delete all terms from the user database.
--
delete from noun;
delete from verb;

@script2
select * from noun where type = sta;
select * from noun where type = hum;
delete from noun where item = bar1;
select * from noun;
delete from noun where item = foo1;
select * from noun;

-- insert into verb values (c1,c2,c3,c4) as clone of release;
-- insert into verb values (d1,d2) with senses 1,2,3;
-- select * from verb;

-- update verb remove senses 1,4,5,10 from term c1;
-- update verb add senses 12,D96 to term c2;
-- update verb set senses 1,2,3,8,9,13 for term c3,c4;
-- select * from verb;

-- update noun set type = plt where term = foo3,bar3;
-- select * from noun;

-- delete from noun where type = sta;
-- select * from noun;

spool off

-- select * from corelex_noun where type = hum,sta;
quit

insert into noun values (foo1,foo2,foo3)
  as clone of employee;
insert into noun values (bar1,bar2,bar3) with type sta;
select * from noun;
quit


Noun(s) deleted


Verb(s) deleted

     >
Noun(s) inserted


Noun(s) inserted


Noun                Category
----------          ----------
foo1                hum
foo2                hum
foo3                hum
bar1                sta
bar2                sta
bar3                sta


Noun(s) selected

Script "script2" completed.

Noun                Category
----------          ----------
bar1                sta
bar2                sta
bar3                sta


Noun(s) selected


Noun                Category
----------          ----------
foo1                hum
foo2                hum
foo3                hum


Noun(s) selected


Noun(s) deleted


Noun                Category
----------          ----------
foo1                hum
foo1                hum
foo1                hum
bar2                sta
bar3                sta


Noun(s) selected


Noun(s) deleted


Noun                Category
----------          ----------
foo1                hum
foo1                hum
bar2                sta
bar3                sta


Noun(s) selected



---------------------------------------------------------------------
To unsubscribe, e-mail: cg-unsubscribe@...
For additional commands, e-mail: cg-help@...

[John F. Sowa]

Dear Naoya,

I would like to add more to Arun's comments about CLCE.

> Is there any working semantic parser for CLCE?

We had a parser for the old version of CLCE that is on
my web site.  But that version is old (2004), and we have
been extending the system quite a bit.  There is no parser
currently available.

> Are the terms used in CLCE based on any ontology?

The purpose of CLCE is to define ontologies.  It is like
logic -- it has no built-in ontology because anyone can use
it to define their own ontologies.

We are currently in the process of making major extensions
to CLCE, but those extensions are still under development.

Following are some slides that I presented at a conference
on Healthcare Information Technology:

   http://www.jfsowa.com/talks/cl_hitps.pdf

The last page of those slides has a list of references,
including the URLs for Rolf Schwitter's web site for
controlled natural languages.  There is also a pointer
to the web site for Attempto Controlled English (ACE).

You can check those web sites for further information.

John Sowa



---------------------------------------------------------------------
To unsubscribe, e-mail: cg-unsubscribe@...
For additional commands, e-mail: cg-help@...

[naoya_arakawa]

Dear Arun & John,

thank you very much for your clarification and pointers!

> The purpose of CLCE is to define ontologies.  It is like
> logic -- it has no built-in ontology because anyone can use
> it to define their own ontologies.

I see.  As statements in CLCE can be converted into FOL, you can associate them with whatever ontologies you like as long as the ontologies can also be converted to FOL.


The reason why I asked the questions was that I'm writing an NL semantic analyzer that *uses* ontologies to discipline the analysis.
The approach is different from controlled language processing in that it has to deal with all the nasty things like ambiguities.
But once logical expressions are extracted, the rest would be
similar: we can use ontologies and logical inferences.

Best,
--
Naoya Arakawa

---------------------------------------------------------------------
To unsubscribe, e-mail: cg-unsubscribe@...
For additional commands, e-mail: cg-help@...