[Attempto] Interface for Protege

[Norbert E. Fuchs]

On 18 Oct 2010, at 14:27, Dilvan Moreira wrote:

>   So, that would be the state of ACE tools (please correct me):
> 	• APE - rock solid (APE webserver ???)
> 	• RACE - not available, but Beta
> 	• AceRules - Alpha
> 	• AceWiki - Alpha
> 	• ACE View - Protege plugin - Beta 
> 	• OWL verbalizer - ???
> 	• ACE Editor - ???
>   Alpha would be not usable (just demonstration).

Since all ACE tools have been developed in an academic environment, I don't believe that the classifications of the release life cycle of commercial software (http://en.wikipedia.org/wiki/Software_release_life_cycle) aptly apply. Instead it seems better to describe the state of each individual tool.

1. APE: The ACE->DRS part is stable and well regression-tested against 3592 test sentences. But APE contains other parts, e.g. DRS->ACE and DRS->OWL/SWRL
that do not yet cover the complete DRS language, and in DRS->ACE there are some known bugs. Also, the APE tokenizer has a known bug which has been fixed but not
released yet. If "APE webserver" is supposed to mean "the APE web-interface calling the APE web-service that runs on the Attempto server", then this should be considered a demo. While it is stable and we do our best to keep it running, we do not guarantee that it is running 24/7. Users are expected to run their own servers/services. Furthermore, the APE web-interface implements an approach to ACE where users are expected to have learned ACE, to submit their ACE text to APE, and to correct any errors that APE finds. Alternatively, the predictive ACE Editor implements an approach where users are guided to enter syntactically correct texts that are then submitted to APE.

2. RACE is currently only available via its web-interface calling its web-service on the Attempto server. This version of RACE does not yet cover all deductions possible in the FOL subset of ACE, and – for reasons that we do not understand – is one order of magnitude slower than the same version on my local machine. I am currently revising RACE for increased speed and scalability, and will extend it to cover the complete FOL subset of ACE. Also, I am considering using the ACE Editor for input. RACE will hopefully be publicly available by the end of the year.

3. AceRules is a stable research prototype. It works fine but hasn't been tested much.

4. AceWiki runs in a stable way and its important parts are tested, but a few features are missing to make it practically useful.

5. Ace View: There are currently not enough tests in place for ACE View, so we cannot confidently estimate how stable it is.

6. The OWL verbalizer at http://code.google.com/p/owlverbalizer/source/detail?r=2 is stable. It is planned to add more features though, e.g. extend the coverage to a fragment of SWRL. The OWL verbalizer web-service that runs on the Attempto server, i.e. the one used by the APE web-interface, is just a demo.

7. ACE Editor is in a stable version though it does not yet cover all of ACE.


On 18 Oct 2010, at 14:27, Dilvan Moreira wrote:

> On 16 Oct 2010, at 16:22, Norbert E. Fuchs wrote:
> 
>> As Kaarel already wrote RACE is not an OWL/SWRL reasoner, but a full FOL reasoner. 
> 
> But wouldn't a FOL reasoner be more general then a DL one (in the sense that it will compute any DL construct)? So if I feed OWL/SWRL assertions (remembering that OWL+SWRL is beyond DL) to RACE wouldn't it return meaningful answers/assertions?

Of course, RACE can do deductions for the ACE subset that can be translated into OWL/SWRL. However, since OWL uses a conceptualisation based on individuals, properties and classes, the results that you expect from the deductions are different from the ones that RACE currently generates. As Tobias Kuhn wrote on 30 Nov 2009 in this mailing list

>> There are several reasons for the fact that AceWiki uses Pellet instead of RACE (but we have some plans to enable the use of RACE within AceWiki in the future).
>> 
>> First of all, RACE does not have a Java interface at the moment, which makes it harder to integrate it into AceWiki.
>> 
>> Furthermore, performance would (potentially) suffer when using a larger subset of ACE (as the one supported by RACE) than the OWL subset (as supported by Pellet). I agree that it would be desirable that AceWiki can do reasoning within a larger subset of ACE than the one supported at the moment. But already now, AceWiki can get quite slow when the knowledge base gets large. And switching to RACE would enlarge this problem.
>> 
>> Another problem is that RACE does not provide (at the moment) an API for retrieving the set of instances for a given class, or for extracting the class hierarchy.
>> 
>> Long answer short: AceWiki could in principle use RACE but it uses Pellet at the moment because of performance and interface reasons.

Tobias and I will come back to this issue once the new version of RACE will be available.


> By the way, APE and RACE are written in Prolog, do they run in some Prolog that runs on the Java JVM? For instance, GNU Prolog fo Java?

We haven't tested this. SWI Prolog that we use for APE, RACE and other tools has a bidirectional interface to Java (http://www.swi-prolog.org/packages/jpl/).

Regards.

  --- nef