Abstract
Protégé OWL1 is an open source tool created to support ontology development for the Semantic Web. It is a plug-in extension to the Protégé ontology development platform. Protégé OWL allows users to edit ontologies in the Web Ontology Language (OWL) and to use description logic classifiers to maintain consistency of their ontologies. Protégé OWL can also assist developers of intelligent applications in biomedicine, because many of the problem-solving tasks they seek to automate can be construed as classification tasks. Protégé OWL provides access to emerging knowledge representation standards such as OWL and high-performance classifiers. Being integrated with Protégé, the OWL Plug-in allows users to exploit Protégé’s core features and services such as graphical user interfaces, a variety of storage formats, and data acquisition and visualization tools. Finally, Protégé OWL provides an API allowing it to be integrated into applications.
Developers of intelligent applications in biomedicine face challenges in representing, managing, sharing, and reusing the knowledge required by their systems. Many reasoning systems use problem-solving approaches that are computationally inefficient because of complex reasoning methods. Such systems are difficult to build and maintain because knowledge is contained in domain ontologies as well as in the application code. Comprehensive methodologies are needed that integrate the domain and reasoning knowledge.
OWL has emerged as a standard language for representing knowledge in the Semantic Web. OWL is also based on description logics, and it supports automated reasoning. If an intelligent application is amenable to being posed as a classification task, then OWL provides the advantage of a standard knowledge representation language that can encode both the domain knowledge as well as reasoning knowledge in the form of axioms and class definitions.
Protégé OWL has been successfully deployed for the last two years. It is implemented in Java, and it runs on a broad range of hardware platforms. Protégé OWL is implemented in Java, and it runs on a broad range of hardware platforms. has an extremely active community of hundreds of users, and it is becoming the de-facto standard OWL editor.
Protégé OWL provides a variety of features that makes it very useful for building ontologies in OWL and intelligent applications that use those ontologies.
Graphical user interface (GUI) and API . Protégé OWL is built upon the Protégé frame-based knowledge model and uses the Protégé GUI for editing classes, slots (properties), and instances (individuals). It provides an API allowing developers to integrate Protégé OWL in their applications.
Graphical editor for Logical OWL Expressions . Protégé OWL provides a comfortable expression editor that allows users to quickly assemble expressions with the mouse or keyboard. It also uses a graphical object-oriented display of primitive and defined classes. The editor supports drag/drop and copy/paste.
Wizards to streamline complex tasks . There are wizards to support common ontology-engineering patterns, such as creating groups of classes, making a set of classes disjoint, creating a matrix of properties in order to set many property values, and creating value partitions.
Direct access to reasoners . Protégé OWL provides direct access to high-performance classifiers such as Racer. The user interface supports three types of reasoning: (1) consistency checking, (2) classification (subsumption), and (3) instance classification).
Because Protégé OWL builds upon the Protégé system the following useful features are also available:
Form generation. Protégé OWL can automatically generate a user interface to acquire instance data from class definitions, a useful feature supporting knowledge acquisition.
Multi-user support. Protégé OWL provides multiuser support for synchronous knowledge entry.
Support for multiple storage formats . Protégé OWL can be extended with back-ends for alternative file formats. Current formats include Clips, XML, RDF, and OWL.
We have used Protégé OWL to develop reasoning applications in two different biomedical domains: inferring penetrating injuries and TNM tumor classification. It may be similarly useful in creating other intelligent applications in biomedicine.