Blog post on the meaning of terms

Hello again,

Following our call on Friday, I added this post to the XDMDL wiki blog, to which you are warmly invited to contribute (or use the comments section to reply to my post).

http://wiki.saltis.org/display/XDMDL/2015/03/29/The+meaning+of+terms

On Call-02, we discussed the disambiguation of terms, e.g. by using Wordnet, particularly with reference to xAPI verbs. As this goes to the heart of what we are trying to achieve (supporting better semantic interoperability) I had some further thoughts about the problem.
My gut feeling is that the sort of global definition that might be provided by something like Wordnet is not going to be sufficient for our needs, even though it might have an important contribution to make. I thought it might be useful to consider the problem in the light of a couple of terms that caused some problems in the SCORM context: "complete" and "finish".
My understanding of the difference is that "complete" was equivalent to "progress=100%" - you had done all that you were expected to do. While "finish" means that the attempt was closed - you had done all you were *going* to do, even it that was not all that you might have been expected to do.
In fact, the distinction is perhaps not quite so clear. Complete does not mean "doing all that there *is* to do" If the activity is a bit like a maze, you would not be expected to walk every inch of the maze before you could be said to have completed it - you would just be expected to get to the middle and then to find the exit. This is equivalent to "meeting your objectives" or "passing" or "mastering" the activity.
But if the activity was a test out of 10, then you might be able to meet your objectives by achieving the pass mark by getting 6 correct answers, which could be achieved by answering only 6 questions and not the full 10.
So the difference between "complete" and "finish" lies not so much in the intrinsic meaning of he verbs, but in the subject - what you are held to be completing or finishing. And there are several possibilities:
1. the set of all interactions that exists within the activity (e.g. to answer all the questions available);
2. a subset of the interactions that are deemed to represent completion (e.g. to answer at least 10 out of the 20 questions available);
3. a subset of correct responses (i.e. to meet preset objectives - find your way out of the maze);
4. the attempt;
5. the session.
In practice, it is left to the Activity Provider to determine what constitutes completion and what constitutes maximum performance, and what the relationship is between them. So what matters is not what, globally speaking, the difference is between "completing" and "mastering", but *who decides* what the relationship is and *what* you are being judged to have completed or mastered.
In SCORM, it is established within the specification how performance is quantified (by raw or scaled scores), it is left to the LMS to decide what the pass mark is, and any relationship between completion and passing is not exposed (i.e. if the AP wants to adjust its measurement of completion in relation to the pass-mark that it has been given by the LMS, it may do so privately.
What this amounts to is a set of:
(a) transactional rules, determining which services are responsible for interpreting which values;
(b) relational rules, explaining how "complete" relates to "pass" or to "finish".
Below are the alternative definitions of "complete" given by Wordnet. Definition 1 corresponds says that "complete" means the same as "finish" but the examples show that this is meant in the sense that SCORM uses the term "complete" and NOT the term "finish" - so the definition is not of much help.
Definition 4 defines "complete" in the same sense that SCORM might use "pass" or "mastered" - so in this case, the Wordnet definitions *could* be used for disambiguation. But even in the case of "complete" and "pass" - would the dictionary definitions be as helpful as a set of rules which defined the terms with reference to the other terms / data values that were being used in that particular data model?
* S: <http://wordnetweb.princeton.edu/perl/webwn?o2=&o0=1&o8=1&o1=1&o7=&o5=&o9=&o6=&o3=&o4=&s=complete&i=0&h=0000000000#c> (v) complete,finish <http://wordnetweb.princeton.edu/perl/webwn?o2=&o0=1&o8=1&o1=1&o7=&o5=&o9=&o6=&o3=&o4=&s=finish>(come or bring to a finish or an end)"He finished the dishes"; "She completed the requirements for her Master's Degree"; "The fastest runner finished the race in just over 2 hours; others finished in over 4 hours"
* S: <http://wordnetweb.princeton.edu/perl/webwn?o2=&o0=1&o8=1&o1=1&o7=&o5=&o9=&o6=&o3=&o4=&s=complete&i=1&h=0000000000#c> (v) complete(bring to a whole, with all the necessary parts or elements)"A child would complete the family"
* S: <http://wordnetweb.princeton.edu/perl/webwn?o2=&o0=1&o8=1&o1=1&o7=&o5=&o9=&o6=&o3=&o4=&s=complete&i=2&h=0000000000#c> (v) dispatch <http://wordnetweb.princeton.edu/perl/webwn?o2=&o0=1&o8=1&o1=1&o7=&o5=&o9=&o6=&o3=&o4=&s=dispatch>,discharge <http://wordnetweb.princeton.edu/perl/webwn?o2=&o0=1&o8=1&o1=1&o7=&o5=&o9=&o6=&o3=&o4=&s=discharge>,complete(complete or carry out)"discharge one's duties"
* S: <http://wordnetweb.princeton.edu/perl/webwn?o2=&o0=1&o8=1&o1=1&o7=&o5=&o9=&o6=&o3=&o4=&s=complete&i=3&h=0000000000#c> (v) complete,nail <http://wordnetweb.princeton.edu/perl/webwn?o2=&o0=1&o8=1&o1=1&o7=&o5=&o9=&o6=&o3=&o4=&s=nail>(complete a pass)
* S: <http://wordnetweb.princeton.edu/perl/webwn?o2=&o0=1&o8=1&o1=1&o7=&o5=&o9=&o6=&o3=&o4=&s=complete&i=4&h=0000000000#c> (v) complete,fill out <http://wordnetweb.princeton.edu/perl/webwn?o2=&o0=1&o8=1&o1=1&o7=&o5=&o9=&o6=&o3=&o4=&s=fill+out>,fill in <http://wordnetweb.princeton.edu/perl/webwn?o2=&o0=1&o8=1&o1=1&o7=&o5=&o9=&o6=&o3=&o4=&s=fill+in>,make out <http://wordnetweb.princeton.edu/perl/webwn?o2=&o0=1&o8=1&o1=1&o7=&o5=&o9=&o6=&o3=&o4=&s=make+out>(write all the required information onto a form)"fill out this questionnaire, please!"; "make out a form"
Hope these musings might usefully inform our future discussions about the relationship between web ontologies and rules-based systems.Best,
Crispin