[Attempto] Using Attempto as process ontology language

[Norbert E. Fuchs]

On 14 Jul 2013, at 13:56 , Edmon Begoli <ebegoli at gmail.com> wrote:

> Dear colleagues,
> 
> I am wanting to use Attempto and related tools as process ontology language for my domain of research (Applied Behavior Analysis/ABA) and for encoding of ABA procedures and rules.
> 
> I would need to encode statements like:
> 
> Behavior is any action that student can undertake.
> 
> Expected behaviors are ... (list here).
> 
> Behavior is followed by a consequence.
> 
> Consequence can be reward or punishment. 
> 
> Reward can be a positive or negative reinforcement.
> 
> Reinforce expected behavior.
> 
> Reinforcement uses reinforcers. 
> 
> Reinforcers are cracker, candy, hug, praise, toy, smile.
> 
> 
> Would you tell me please what are the steps i would need to take (lexicon building, etc.) to leverage Attempto tools for my task, including using RACE for consistency checking,
> 
> Thank you,
> Edmon


Edmon

Interesting application of Attempto Controlled English and its tools.

Here are my answers to your questions.

1. See the ACE Lexicon Specification at http://attempto.ifi.uzh.ch/site/docs/ace_lexicon.html that describes the format of the user lexicon that takes preference over APE's built-in lexicon. Check the ACE documentation  at http://attempto.ifi.uzh.ch/site/docs/ for details of the language ACE.

2. RACE: I am about to finish the implementation of reasoning with arithmetic and will then – after some clean-up – submit RACE to GitHub.

3. I ACE'ified your example sentences using the APE web-interface to reformulate them. For simplicity I did not define a user lexicon but preceded unknown words like "behaviour" and "reinforce" by a prefix that defines their word class, e.g. "n:behaviour", "v:reinforce" (see the APE help text). 

One general remark: All ACE nouns need a determiner.

> Behavior is any action that student can undertake.

Every behaviour is an action that a student can undertake. 
(Note that I used the universal quantifier "every" instead of the existential quantifier "a".)

> Expected behaviors are ... (list here).

The expected behaviours are X and Y and Z. 
(I replaced your undefined list just by some bare variables.)

> Behavior is followed by a consequence.

Every behaviour is followed by a consequence. 
(Note that I used the universal quantifier "every" instead of the existential quantifier "a".)

> Consequence can be reward or punishment. 

A consequence can be a reward or can be a punishment. 
(Here I am not sure whether you mean "every consequence" or "a consequence". Note that in ACE the disjunction "or" cannot combine noun phrases, but only verb phrases and sentences. Thus I repeated the "can be".)

> Reward can be a positive or negative reinforcement.

A reward can be a positive reinforcement or can be a negative reinforcement. 
(Again I am not sure whether you mean "every reward" or "a reward". See my note above concerning "or" and noun phrases. Further note that in ACE the disjunction "or" cannot combine adjectives, and that eliding the noun phrase is not allowed. Thus I introduced two complete verb phrases.)

> Reinforce expected behavior.

X, reinforce the expected behaviour!
(This I interpreted as a command. In ACE a command needs an addressee – here I introduced the variable "X" – and an exclamation mark. Note that RACE does not process commands.)

> Reinforcement uses reinforcers. 

Every reinforcement uses some reinforcers.

> Reinforcers are cracker, candy, hug, praise, toy, smile.

Every reinforcer is a cracker or is a candy or is a hug or is a praise or is a toy or is a smile.
(I interpreted your list as disjoined by "or", in which case complete verb phrases are required. For the non-probable case that the list is meant as conjunction, you can conjoin noun phrases by "and".)


Let me know how this works out for your project.

Best regards.

Norbert E. Fuchs
Department of Informatics & Institute of Computational Linguistics
University of Zurich