Re: [ontology-summit] First Model Bench Challenge

[Simon Spero <sesuncedu@xxxxxxxxx>]

On Thu, May 3, 2012 at 10:41 AM, henson graves <henson.graves@xxxxxxxxxxx> wrote:

The purpose of the Model Challenge problems is to assess the capability of modeling languages and/or logics to represent various constructions which occur in a variety of conceptual modeling disciplines. The first several challenges are concerned with what is called composite structures (See Cory’s site). In the marriage example the composite structure is the husband and wife components of the marriage and the spousal relationship. Again the problem is to create model which describes one or more marriages.

 

The modeling quality evaluation is precision: are non-intended models are excluded.  This requires that the language have some implicit or explicit notion of semantics. This problem is simplified to only be concerned with the static case. Later temporal aspects will be investigated.

It is in general not considered wise to use precision on its own as a metric for evaluating systems; this is due to the ability to arbitrarily boost precision by reducing recall. As a long as a single relevant item is selected, and no non-relevant items are selected, precision becomes 1, regardless of how many relevant items may exist.  

For  example, under CWA, the following representation (in subset of ACE), has perfect precision; however, the recall is almost zero. [ESOM is short for EuropeanStatisticalOfficeCompliantMarriage]

"There is exactly 1 ESOM and it hasHusband Alex and hasSpouse him and hasWife Emily and hasSpouse her. There is exactly 1 man and he is Alex and isHusbandOf Emily and isSpouseOf her. There is exactly 1 woman and she is Emily and isWifeOf Alex and isSpouseOf him."   

This is not especially useful... 

-----

Most  of the ESOM constraints are easily captured in OWL; others are not as well supported.  For example, we can specify a role chain so saying that the the wife in the ESOM that something is the husband in  is the wife of the husband, but we cannot state that the two are equivalent.  Asserting that w isWifeOf h does not  entail the existence of an ESOM e. 

Simon

_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/   
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/  
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2012/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2012  
Community Portal: http://ontolog.cim3.net/wiki/     (01)