Readme file to accompany the Protege import versions of the SUMO ontology, plus the SUMO supplemented with the Quality of Service ontology (from Adam Pease and Teknowledge) and the Core Components mappings created by Adam Pease Nov. 2004. This informationn is current as of November 21, 2004. The Protege ontology containing the SUMO and Mid-Level combined ontologies (KIF format is file SAM156b.txt), plus the Quality of Service ontology (KIF file QOSONT.KIF), plus the Core Components classes suggested by Adam Pease, (collected in the KIF file CCTrep2.txt) is contained in the three Protege files : SMQCC.pprj SMQCC.pins SMQCC.pont This ontology was constructed by adding the concepts in the Core Components mappings provided by Adam Pease (in KIF file CCTrep2.txt) to the SMQ ontology. The SMQ ontology contains the SUMO and Mid-Level combined ontologies (KIF format is file SAM156b.txt), plus the Quality of Service ontology (KIF file QOSONT.KIF). The KIF-format classes in file CCTrep2.txt correspond logically to the KIF-format classes provided by Adam Pease, but in some cases the format of the classes and relations provided by Adam Pease was reorganized to conform to the format expected by the SkifTab import program, mostly by adding a standard documentaion propostion to each subclass declaration.. The , or by changing some apparent typographical errors. The base Protege ontology containing the SUMO and Mid-Level combined ontologies (KIF format is file SAM156b.txt), which was used to import these concepts is called SMQ and is contained in the three Protege files : SMQ.pprj SMQ.pins SMQ.pont The SMQ ontology was constructed by adding the concepts in the Quality of Service ontology provided by Adam Pease (KIF file QOSONT.KIF) to the SAM156bX3 ontology, which contains the concepts in the combined SUMO and Mid-Level ontologies. The Protege ontology containing the SUMO and Mid-Level combined ontologies (KIF format is file SAM156b.txt), is contained in the three Protege files : SAM156bX3.pprj SAM156bX3.pins SAM156bX3.pont If any errors or omissions are present in any of these files -- KIF format or Protege format -- please let me know and I will correct the problems. ============================================ NOTE about the import program: The importation of KIF files into Protege is performed by use of a Protege plug-in called SkifTab. The rpesent version of this program ahs been tested only within Protege version 2.1.2, using the Java runt-time environment provided with the Protege download. The java files SkifTab.java and Predicates.java contain the source code for the import program. This program is still crude, but appears at present to convert all of the logical information in the original SKIF-format file into some Protege representation. The axioms are saved only as strings, and higher-arity relations are represented as slots with additional facets -- but instances of higher-arity relatitons need to be reifed as instances of special classes. The import program makes some assumptions about the format of the SKIF file, with the result that a logically correct SKIF file in which the propositions are not in a particular order will give errors during the importing process. Thus reorganization of a SKIF file may be required to get the importation to proceed correctly. The most important format assumptions are: (1) all propositions with a specific constant in argument 1 position must be collected in a single block of sequential lines, and separated from other blocks by one or more blank lines. (2) class definition blocks will have a subclass proposition as the first proposition (3) relation definition blocks will have either an instance proposition or a subrelation proposition as the first proposition of the block (4) instance definition blocks will have an instance proposition as the first proposition of the block (5) all blocks must have a documentation proposition as the final proposition, unless they are bounded by special tags (6) every proposition other than a documentation proposition must be a single line This was the format followed by the original SUMO SKIF file, but it is not followed in all subsequent ontologies provided by Teknowledge. Some reorganization of those SKIF files is thus usually necessary for the import process. The import program can be made more flexible, but it was decided that requiring a specific format is better, because certain types of errors in the import files are easier to detect when they are organized in this manner. At present the program cannot correctly export all information from a Protege file to a SKIF-format file, and this program is therefore usable at present onyl as a visualizing tool, not for development or editing. Recently an additional problem has arisen. Under certain circumstances (frequently, recently), an internal Java interpreter error occurs, in which the program exits without any useful error information. This is apprarently a bug in the optiization routine of the Java run-time program, and can be avoided by runninng the program in interpret-only (no optimization) mode, at the cost of a substantially slower execution. I do not know at present whether later versions of Protege will manage to avoid the conditions that cause this bug to surface. Pat Cassidy cassidy@micra.com