The goal of this set of pages is to identify and recommend a set of guidelines that contribute directly to quality, or indirectly by pursuing ontology reusability and adoption.
The following text itself is still subject to finetuning
Developing good quality ontologies is an important goal in ontology engineering. The importance of quality is not only to build reliable ontologies, but also good quality enables reusability, consensus, adoption, correctness in reasoning and prediction, good performance in computation, etc. However, achieving an agreed upon or a principles-based criteria set that can be generalized to assess ontologies remains a very difficult task.
These guidelines will be promoted not only to ontology engineers but also to tool developers. An ontology authoring tool can then be evaluated and scored, for example, based on how much it implements these guidelines. The idea is that enforcing these guidelines during the ontology development phases ensures a certain quality of the product, i.e. the ontology. Although the final recommendations are not intended to play a role of a gold standard for quality assessment, but as a first initiative in this regard, they are supposed to lead to better ontology content authoring.
Experienced ontologists are invited and encouraged to contribute to this recommendation, by submitting guidelines based on their best practice and research findings. A guideline is not necessarily a rigid assessment criteria or a theory, but can also be a methodological recommendation that guides ontology builders to achieve better quality, reusability and/or adoption. Not every guideline includes a formal technical specification, nor can every recommendation be embodied by support from authoring tools; each is intended to be directly comprehensible by industrial users and applicable in a wide range of industry settings. While some guidelines are interrelated, each is well-contained and can be followed independently of the others.
The collected guidelines will be refined to arrive at more agreement; where there is eventual disagreement, the differing opinions will be clearly documented. All guidelines will be reformulated where appropriate to provide a coherent approach to ontology content. Each guideline should be easy to understand and apply by normal ontology engineers or non-technical domain experts.
Within the OOA web site, ontology authoring quality guidelines will follow the template laid out in the first guideline. By copying this template and filling it in, relatively uniform guideline pages will appear that can easily be collated and harmonised later on.
Experienced ontologists are invited and encouraged to contribute to this recommendation, by submitting guidelines based on their best practice and research findings. A guideline is not necessarily a rigid assessment criterion or a theory, but can also be a methodological recommendation that guides ontology builders to achieve better quality, reusability and/or adoption. Not every guideline includes a formal technical specification, nor can every recommendation be embodied by support from authoring tools; each is intended to be directly comprehensible by industrial users and applicable in a wide range of industry settings. While some guidelines are interrelated, each is self-contained and can be followed independently of the others.
The collected guidelines will be refined to arrive at more agreement. Where there is eventual disagreement, the differing opinions will be clearly documented. All guidelines will be reformulated where appropriate to provide a coherent approach to ontology content. Each guideline should be easy to understand and apply by normal ontology engineers or non-technical domain experts.
Guideline collection will be performed online, using the web site of the OOA as platform. Specific guideline templates have been designed and provided in the content management system, and a review procedure has been set up as a work flow to allow a few cycles of comments (Delphi method) before a guideline will be released as ‘formally recommended’. However this procedure will not be closed until formal release. The transient status of a not-yet-released guideline will be made clear, but since the whole idea is to get comments and feedback, even preliminary guidelines will be accessible. The process will closely follow the well-known and established RFC process of the Internet Engineering Task Force. This model calls for an open submission using quite strict format and procedure requirements, a review and revision by a small group of renowned specialists, and a cycle of publish-and-feedback. The OOA plans to have a continuous process in place, to avoid an unnecessary long interval between submission and approval/dissemination of a new guideline.