[Attempto] data properties

[Kenneth Jones]

Not only would I be willing to contribute, but I am actually working right
now on some code that I plan to offer to the Attempto group.  It doesn't
matter to me whether the code is hosted on the Attempto site or on
SourceForge.

My current focus is on developing a platform for the average programmer
without a CS degree and for non-programmer scientists, engineers, and
business analysts who are able to express their needs in declarative form,
can handle concepts like transitivity, and are reasonably good at making and
understanding valid deductions.  In other words, I am targeting an audience
that is a bit more comfortable dealing with reasonably simple formal systems
(though they may not think of them as "formal systems") than the current
target audience of AceWiki.  They would only need to be capable of informal
logical deduction and would need no knowledge or programming, formal logics,
tractability, linguistics, etc.

To this end I have developed a partial implementation of a parser
(implemented in Java) for the Prolog-based textual representation of
Attempto DRS version 6.5 that converts the text into a parse tree of Java
objects.  I have also partially completed Java code that converts the parse
tree into an "abstract syntax network" of sorts that consists of the graph
of Java objects that results from resolving all the textual references in
the DRS into links in the network (edges in the graph are implemented as
field references between the Java objects that form the nodes of the
network).  I am beginning work on a transformation from the abstract syntax
graph to Java code that performs inferences based on the structure of the
DRS.  Simple Java function calls will be used to query the system.  I also
hope to develop a simulation capability that can be used prior to code
generation.

In the beginning I envision three applications.  Users would start with an
Ace Editor-based app that provides predictive editing and Java-based
question answering.  After learning basic ACE the user would be able to move
to the more powerful capabilities of an APE-based app that provides neither
predictive editing nor inference capabilities, but allows a higher level of
expressibility, or to an AceWiki-based app that trades expressibility for
tractability, uses OWL and a DL reasoner, and provides all the extras that
AceWiki has to offer.  Eventually I would like to see a spectrum of apps or
an integrated system with a spectrum of capabilities, allowing the user to
trade expressibility for inferential capabilities and computational
complexity.

This is a very ambitious plan and the devil is in the details.  I am not
sure all my ideas will pan out and the time I have available for working on
this project is limited, but I am spending virtually all my free time on
it.  If I am able to develop anything that the Attempto group would find
useful I would be happy to contribute it.  I would also be open to working
on other Attempto-related projects.

-- Ken

On Tue, Dec 22, 2009 at 3:10 AM, Tobias Kuhn <t.kuhn at gmx.ch> wrote:

> Ken,
>
> Agreed! I put it to the top of my todo list. (But it is not an easy change,
> since it requires significant changes in the grammar and probably also in
> the predictive editor.)
>
> I will be abroad for the first half of 2010, however, and I won't be able
> to work on AceWiki all the time. In my view, the best situation would be if
> a small "AceWiki community" emerged so that I could outsource part of the
> AceWiki development work into the community. See my SourceForge suggestion
> in the other message.
>
> Maybe it's a good moment to ask: Who would be able and willing to
> contribute to a public AceWiki code base, and thereby becoming a member of a
> new AceWiki community?
>
>
> Cheers,
> Tobias
>
>
>
>
> On 22.12.09 05:37, Kenneth Jones wrote:
>
>> Tobias,
>>
>> I would just like to follow up on George's question about OWL data
>> properties by stating that I would like to see support for strings and
>> numbers moved to the top of your list of priorities.  Numerical data is
>> an essential element of most scientific, engineering, and business
>> domains and string handling is an important part of many business
>> applications.  I think that support for numbers is more important than
>> support for strings, but I am guessing that developing support for
>> strings would be so similar to developing support for numbers that
>> developing support for both would require only a small amount of
>> additional effort after support for one had been implemented.  I think
>> data property support is a fundamental requirement for a number of
>> real-world domains and I would like to see AceWiki's capabilities
>> extended in ways that will make it a viable platform for a wider range
>> of applications.
>>
>> -- Ken
>>
>>
>>
>> _______________________________________________
>> attempto mailing list
>> attempto at lists.ifi.uzh.ch
>> https://lists.ifi.uzh.ch/listinfo/attempto
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ifi.uzh.ch/pipermail/attempto/attachments/20091223/be22574a/attachment.htm>