Semantic Services for Information and Management Support in Mass Casualty Incident Scenarios Uwe Krüger, Aygul Gabdulkhakova, Birgitta König-Ries, and Clemens Beckstein Institute of Computer Science, Friedrich-Schiller-University Jena, Germany {uwe.krueger,aygul.gabdulkhakova,birgitta.koenig-ries, clemens.beckstein}@uni-jena.de http://www.speedup.uni-jena.de Abstract. Operation managers in mass casualty incidents (MCI) are easily overwhelmed by the highly dynamic scenario. Today, they are often supported by paper-based checklists that help ensure that nothing important gets overlooked. In this paper, we present our approach to dramatically increase the support offered by such checklists: We suggest to replace them by IT-based intelligent checklists that will not only display information the manager should collect and tasks he might perform, but directly support semi-automatic information gathering, propagation and task-execution. Keywords: checklists, semantic services, service composition. 1 Introduction In large scale scenarios up to mass casualty incidents (MCI), the rescue forces at the scene are overwhelmed by the number and severity of casualties. In order to successfully handle MCIs, operation managers in situ have to manage the gap between available and necessary resources [1]. In regional rescue coordination centers special plans and checklists are used to guide through lists of tasks to be done, because in stressful situations such lists prevent dispatchers from forgetting important steps. But checklists are also useful for operation managers in the field — if certain items are skipped or missed, the result can be harmful or even fatal. Let us look at an example of a classical paper-based checklist containing the following entry addressed to a MCI manager: “If hazardous substance is involved, inform all leaders about this fact! ”. This entry does not specify who the addressee of this information are and how they can be informed about the danger. These decisions can only be made in the specific MCI situation where hazardous substance was detected and therefore are left to the manager — however busy and stressed he may be in such a dynamic situation. First investigations of real MCI incidents have shown that an electronic assistant, in the form of an intelligent checklist handler, taking over part of this responsibility, allows the manager to M. Cezon and Y. Wolfsthal (Eds.): ServiceWave 2010 Workshops, LNCS 6569, pp. 59–66, 2011. c Springer-Verlag Berlin Heidelberg 2011  60 U. Krüger et al. better focus on the core problems which tremendously improves the MCI management process. This is also confirmed by the requirements studies [2, 3]. In this paper we present the concept of an intelligent checklist support for the management of MCIs. This work is part of the SpeedUp1 project which aims to build up an information and communication system for the operations managers. Section 2 introduces intelligent checklists and explains how they can be used in MCI management processes. Section 3 develops our proposal for an architecture that allows to build intelligent electronic checklist support systems (IECSS). This architecture is related to other approaches to the problem at hand in Section 4. The paper concludes with a summary and a discussion of the role that this architecture plays in the overall SpeedUp project. 2 From Checklists to Intelligent Checklists In the MCI area checklists typically are not just simple verifiable questions. Rather, they are more or less abstract descriptions of tasks that govern how humans act and communicate in the MCI process. These tasks often leave open important details which are necessary for their successful completion because they can only be realized with the full knowledge of the specific situation at hand. MCI-checklist entries are of a dual nature: on the one hand they describe a task to be done and on the other the goal that the task is to accomplish. These entries usually are given as natural language orders to the checklist addressee. In order to be usable in an IECSS they therefore first have to be formalized: the semantics (correct usage) of the concepts addressed in the entries (their ontology) has to be fixed and made available for the formulation of pre- and postconditions that control the interplay of the tasks and specify the effects (goals) that the task execution has (should have) in the real world. Checklists for MCIs typically exhibit a hierarchical structure. This structure may be given explicitly by a higher level checklist entry referring to a lower level checklist implementing this entry or just implicitly by an abstractly formulated entry where the details of its (situation specific) implementation are left to the checklist user. In the SpeedUp project these implicit relationships were made explicit and formalized as hierarchical task networks (HTNs) via a detailed analysis of the strategies used by professionals to handle complex MCIs. Depending on the current situation a suitable subset of these HTNs can therefore now be selected either by man or machine to control how abstract tasks are realized. In contrast to classical paper-based checklists, entries in MCI-checklists need and often must not be executed in a rigid a priori given order. Different tasks corresponding to different entries of the same checklist and even entire checklists typically are pursued concurrently and therefore hard to manage. The user has to continuously monitor and orchestrate their execution according to the dependences between the tasks involved. This is also the primary design goal for our intelligent electronic checklist support system. 1 SpeedUp is funded within the Federal Government’s program Research for Civil Security by the German Federal Ministry of Education and Research.