Edhibou: a Customizable Interface for Decision Support in a Semantic Portal

arXiv:0811.0310v1 [cs.AI] 3 Nov 2008 EdHibou: a Customizable Interface for Decision Support in a Semantic Portal Fadi Badra LORIA (UMR 7503 CNRS– INPL–INRIA-Nancy 2–UHP) Vandœuvre-lès-Nancy, France badra@loria.fr Mathieu d’Aquin Knowledge Media Institute The Open University United Kingdom m.daquin@open.ac.uk Jean Lieber LORIA (UMR 7503 CNRS– INPL–INRIA-Nancy 2–UHP) Vandœuvre-lès-Nancy, France lieber@loria.fr Thomas Meilender LORIA (UMR 7503 CNRS– INPL–INRIA-Nancy 2–UHP) Vandœuvre-lès-Nancy, France meilendt@loria.fr ABSTRACT The Semantic Web is becoming more and more a reality, as the required technologies have reached an appropriate level of maturity. However, at this stage, it is important to provide tools facilitating the use and deployment of these technolo- gies by end-users. In this paper, we describe EdHibou, an automatically generated, ontology-based graphical user in- terface that integrates in a semantic portal. The particularity of EdHibou is that it makes use of OWL reasoning capabili- ties to provide intelligent features, such as decision support, upon the underlying ontology. We present an application of EdHibou to medical decision support based on a formaliza- tion of clinical guidelines in OWL and show how it can be customized thanks to an ontology of graphical components. 1. INTRODUCTION The Kasimir project is a multidisciplinary project which aims at providing oncology practitioners of the Lorraine region of France with decision support and knowledge manage- ment tools. The Kasimir system is a clinical decision sup- port system which relies on the formalization of a set of clinical guidelines issued by the regional health network. It uses decision knowledge contained in an OWL ontology to provide decision support to clinicians. In such an ontology O, a class Patient denotes the class of all patients, a class Treatment denotes the class of all treatments and a property recommendation links a class of patients to a class of recom- mended treatments. Then to a class P of patients is associated a treatment T by an axiom P ⊑∃recommendation.T where (P ⊑Patient T ⊑Treatment (1) A medical situation is representedby an instance a of the class Patient in the ontology O. The system then exploits axioms of the form (1) to associate a set of recommended treatments to the patient represented by a. Deciding which treatments to recommend to the patient represented by a amounts to finding the most specific atomic concepts T in O such that ⊨O (∃recommendation.T)(a) holds. The original motivation when developing EdHibou was to provide a user interface for the Kasimir system that lets the user describe a medical situation for which a decision has to be taken. Such a graphical user interface should let the user complete the description of an OWL instance a and trigger some reasoning tasks on the underlying OWL representation of clinical guidelines to propose a set of recommended treat- ments. We built EdHibou as a generic framework, allowing application developers to generate customizable interfaces to ontologies and ontology reasoning. The Kasimir system takes advantage of this framework as an application of Ed- Hibou. The key idea in EdHibou is to allow the end-user to edit an OWL instance without having to manipulate the OWL syntax, by simply filling in values in a form. When develop- ing this application, the main requirements were to make it generic — so that it can be easily reused in other applications, and easy to deploy. It also had to be customizable. 2. SYSTEM ARCHITECTURE Our goal in developing EdHibou was to build a lightweight knowledge edition tool with (1) a very flexible knowledge model, and (2) highly configurable knowledge acquisition forms. Apart from the dynamic user interface update mech- anism, anything had to be configurable, including the choice of the components to display and how they are displayed. Requirement (1) has been met by externalizing the knowl- edge model to a distant knowledge server. The role of this knowledge server is to manage a knowledge base and per- form all reasoning tasks over OWL ontologies. Requirement (2) was fulfilled by pushing application configuration into an ontology. The generation of the user interface is then han- dled by a simple wrapper that takes as input an automatically generated XML representation of the content of an ontology together with a set of graphical component implementations. EdHibou implements a Model-View-Controller architecture pattern (see figure 2) and was developed using the Google Web Toolkit Java AJAX programming framework. K-OWL, the knowledge server, is a standalone component that plays the role of the model. Though it manages knowledge, and not persistent data, K-OWL has been designed in quite the same spirit as standard database management systems. It stores a set of Java models of OWL ontologies that are created with the Jena Java API coupled to the OWL DL reasoner Figure 1: EdHibou’s software architecture. Pellet. These ontologies

…(Full text truncated)…

This content is AI-processed based on ArXiv data.