Given the concepts I've could see in the homepage of Qi4j project, the question arises to me if something of I'm actually doing in the fields of meta-modeling and semantics could help.
For example, in the concept 'Behavior depends on context' I'm actually modeling entities that change their types, an even their names given the actual relationships and roles they participe in. (Person being Employee, Husband, Son, ets. I call this type promotion due to context) X called some way or another given who is metioning it, etc. The actual model and core classes come from an implementation of a type-object, dynamic object model implementation, which derived in this, when I was in the path of building a KR/Ontology management system I've seen many of the options available, and I've thinked there should a way to finest grain decompose and represent information and schema data, providing there will be a way to decompose, for later composing the information in given format should be required. In the middle, this 'abstract representation' will allow to enrich data provided, by merging ontologies, reification, entailment and inferencing. The whole idea consists in treating entities in an ontology as being one of three kinds: Signs (Names), Object (Values) and Concepts (Types). This is a way of 'semantic' interpretation/representation of data-metadata but rather also a 'semiotic' one, in which treatment of signs, objects and meaning relationships is done basically this way. Once we have reached the Type-Name-Object level of abstraction (where models, meta models and meta meta models can reside) the task to be acomplished is to arrange the data in a manner where inference, entailment and reification may occur. This should be done grouping or 'aggregating' the entities this way: Values of the same Name Values of the same Type Names of the same Value Names of the same Type Types of the same Value Types of the same Name The 'aggregation' is done in a Container-Containee abstraction pattern fashion, where the objects contained in a container are inside the container playing an specific Role, and, the container itself, in turn, is containg its children in another given role. This way one could find, given, for example, that many values are inside a type in the same role, that those values are at least 'similar', or also they are the same value, maybe with different name. So, one could find answers, and see graphically where objects and nodes in the graph are the same thing, refer to the same thing, or are logically 'grouped' thogether, conforming a bigger entity. A tool should be developed to manipulate objects like that. The work is the early stages of development, only coded the core classes of the main data structures are coded. There should be necessary later to build a parser, one of a given textual grammar for this model, one for a XML representation of the model, and another tool for generating output in one of these representations from, for example, a RDBMs schema and data or XML/RDFS/OWL datasources. More information available at: https://xama.dev.java.net/ or http://www.nabble.com/file/p25065513/OntXA.zip OntXA.zip Regards, Sebastian Samaruga -- View this message in context: http://www.nabble.com/Data-Metadata-classification-mechanism-tp25065513p25065513.html Sent from the Qi4j-dev mailing list archive at Nabble.com. _______________________________________________ qi4j-dev mailing list [email protected] http://lists.ops4j.org/mailman/listinfo/qi4j-dev
