CP RSS Channel
About Our Sponsors
Articles & Papers
Technology and Society
|STARLab ORM Markup Language (ORM-ML)|
[September 28, 2002] The research team at STARLab (Systems Technology and Applications Research Laboratory, Vrije Universiteit Brussel) has developed an Object Role Modeling markup language (ORM-ML) for representing ORM models in an XML based syntax. Stylesheets may be written to convert this ORM-ML syntax into other syntaxes for processing by business rule engines. Research at the STARLab is "situated in the field of Information Systems, focusing on methodological design of information systems, text retrieval and text enrichment, (semantic) web-based applications, digital libraries, ontologies, integration and cooperation between several Information systems." In its DOGMA project, the team is setting up an ontology server in order to assist the gathering and incremental growth of ontologies.
"ORM (Object Role Modeling) is a method for designing and querying database models at the conceptual level, where the application is described in terms readily understood by users, rather than being recast in terms of implementation data structures. This high-level approach is philosophically in tune with the business rules movement evangelized by such industry leaders as Barbara von Halle and Ron Ross. Typically, a modeler develops an information model by interacting with others who are collectively familiar with the application. Because these subject matter experts need not have technical modeling skills, reliable communication occurs by discussing the application at a conceptual level, using natural language, analyzing the information in simple units, and working with instances (sample populations). ORM is specifically designed to improve this kind of communication. It comes in a variety of flavors, including natural language information analysis method (NIAM), which is best known in Europe, where the method originated in the mid-1970s. Since then, ORM has been extended and refined by researchers in Australia, Europe, the U.S., and elsewhere. Unlike ER, which has dozens of different dialects, ORM has only a few dialects with only minor differences." [from Terry Halpin, "Business Rules and Object Role Modeling."]
The STARLab's XML representation for ORM is described in a document prepared for CoopIS 2002: 'A Conceptual Markup Language that Supports Interoperability between Business Rule Modeling Systems': "The STARLab ORM Markup Language is an XML-based ORM markup language (ORM-ML) with a complete grammar defined in the XML Schema. This markup language enables exchanging ORM models including ORM-syntax business rules. To facilitate validation for example, or just to provide formal and consistent documentation, the team has also developed a verbalization style sheet for ORM-ML documents that allows presenting the facts and the rules in pseudo natural language sentences... The team has chosen ORM for its rich constraint vocabulary and well-defined semantics and elected to use XML Schema to define this communication 'protocol' for conceptual schemas seen as XML document instances. The team has chosen to respect the ORM structure as much as possible by not 'collapsing' it first through the usual relational transformer that comes with most ORM-based tools (or UML, or EER tools for that matter -- after all, these tools were all conceived mainly to build database schemas for in-house use. It is fundamental as well as illustrative of the STARLab approach to emphasize the distinction between ORM-ML itself and the related and interesting work that has been reported in [by Bird, Goodchild, and Halpin] using ORM to design XML document instances, i.e., which contain instance data described in XML using XML Schema language... In the STARLab approach, ORM-ML represents ORM models textually, and the syntax of the resulting model is marked-up by XML tags' syntax (i.e., XML-based structured text document). Therefore the content of this XML document is exactly equivalent to the input ORM model, except for the geometrical information (e.g., shapes, and its positions). The latter could be considered as graphical information of an ORM diagram. The team therefore defined an XML Schema that can act as a grammar to any ORM-ML document... In short, the distinction between ORM-ML, and using ORM to generate XML Schema, is that the output document in ORM-ML is a text representation of the ORM model itself, while in the earlier approach the output document is a transformation from ORM model to an XML Schema instance, which is no longer 'ORM'. The team has designed an ORM Meta schema; its populations are ORM schemas, which the algorithm transforms into ORM-marked-up XML documents... In this approach the ORM diagram's linguistic elements (names of LOTs, NOLOTs, etc.) stay at the level of string values, emphasizing their flexible instance status -- while for Bird, Halpin et al. these names become XML tag names, reducing the flexibility by 'freezing' them in the generated XML model..."
- STARLab website
- DOGMA Ontology research
- "Business Rules and Object Role Modeling." By Terry Halpin. See also the associated ORM website.
- STARLab publications
- "Rule Markup Language (RuleML)."
- "Markup Languages and Semantics."
[September 28, 2002] "A Conceptual Markup Language that Supports Interoperability between Business Rule Modeling Systems." By Jan Demey, Mustafa Jarrar, and Robert Meersman (VUB STARLab, Vrije Universiteit Brussel, Belgium). Paper prepared for CoopIS 2002 (Tenth International Conference on Cooperative Information Systems, October 30 - November 1, 2002, University of California, Irvine). 17 pages. "The Internet creates a strong demand for standardized exchange not only of data itself but especially of data semantics, as this same internet increasingly becomes the carrier of e-business activity (e.g., using web services). One way to achieve this is in the form of communicating 'rich' conceptual schemas. In this paper we adopt the well-known CM technique of ORM, which has a rich complement of business rule specification, and develop ORM-ML, an XML-based markup language for ORM. Clearly domain modeling of this kind will be closely related to work on so-called ontologies and we will briefly discuss the analogies and differences, introducing methodological patterns for designing distributed business models. Since ORM schemas are typically saved as graphical files, we designed a textual representation as a marked-up document in ORM-ML so we can save these ORM schemas in a more machine exchangeable way that suits networked environments. Moreover, we can now write style sheets to convert such schemas into another syntax, e.g., pseudo natural language, a given rule engine's language, first order logic... The ORM conceptual schema methodology is fairly comprehensive in its treatment of many 'practical' or 'standard' business rules and constraint types. Its detailed formal description makes it an interesting candidate to nontrivially illustrate our XML based ORM-markup language as an exchange protocol for representing ORM conceptual models...We describe the main elements of the ORM-ML grammar and demonstrate it using a few selected elementary examples. A complete formal definition of the grammar for this ORM-ML is an XML Schema instance... ORM-ML allows the representation of any ORM schema without loss of information or change in semantics, except for the geometry and topology (graphical layout) of the schema (e.g., location, shapes of the symbols), which we however easily may provide as a separate graphical style sheet to the ORM Schema... Verbalization of a conceptual model is the process of writing its facts and constraints in pseudo natural language sentences, which assumedly allows non-experts to (help) check, validate, or even build conceptual schemas. In ORM-ML, generating such verbalizations from agreed templates (i.e., 'template NL' syntax) parameterized over the ORM schema is done by building separate XML-based style sheets. Moreover, multilingual style sheets also become easier by translating these template sentences into different languages, its parameter values (which come from the ORM schema) translated by a human or machine..." [alt URL; cache]
[September 28, 2002] "A Markup Language for ORM Business Rules." By Jan Demey, Mustafa Jarrar, and Robert Meersman (VUB STARLab, Vrije Universiteit Brussel, Belgium). Paper presented at the International Workshop on Rule Markup Languages for Business Rules on the Semantic Web (14-June-2002, Sardinia, Italy). 21 pages. A later version of the paper is given in CoopIS 2002, "A Conceptual Markup Language that Supports Interoperability between Business Rule Modeling Systems." -- "Conceptual modeling techniques such as EER, ORM and to some extent the UML have been developed in the past for building information systems. These techniques or suitable extensions can often also be used to design business rules at a conceptual level. In particular in this paper we adopt the well-known CM technique of ORM, which has a rich complement of business rule specification, and develop ORM-ML, an XML-based markup language for ORM. Clearly domain modeling of this kind will be closely related to work on so-called ontologies and we will briefly discuss the analogies and differences. Since ORM schemas are typically saved as graphical files, we designed a textual representation as a marked-up document in ORM-ML so we can save these ORM schemas in a more machine exchangeable way over the Internet. Moreover, we can now write style sheets to convert such schemas into another syntax, e.g., pseudo natural language, first order logic, a given rule engine's language, etc. [cache]
|Receive daily news updates from Managing Editor, Robin Cover.|