MLPM: A Multi-Layered Process Model Toward Complete Descriptions of People’s Behaviors Zhang Zuo, Hung-Hsuan Huang, Kyoji Kawagoe Graduate School of Information Science and Engineering Ritsumeikan University Kusatsu, Shiga JAPAN e-mail: {gr0186rk@ed, huang@fc, kawagoe@is}.ritsumei.ac.jp Abstract—Despite the rapid progress in the development of sensor technologies, as well as of information management, no technology exists for recording all the activities of people in the many varieties of human societies. In this paper, we propose a novel process meta model for describing people’s activities that uses a Multi-Layered Process description Model, MLPM. The meta model allows models for various kinds of human social activities, such as sport plays, medical treatment, and agent communications, to be easily described. The significance of this model is that it can be used not only for searching a subpart of people’s activities given a process query, but also for fostering a young novice by presenting to him/her behavior patterns, differentiating between those of the expert and the novice. The meta model can also be used for detecting outliers in process databases. In this paper, we describe the components and structures of the MLPM. It is mentioned that the MLPM is a model suitable toward complete descriptions of people’s behaviors by comparing it with other methods. Keywords—process; behavior; action, meta model; searching. I. I NTRODUCTION Recently, the use of various kinds of sensor devices, such as motion and location sensors, has become widespread. As a result of this rapid popularization, huge amounts of their mon- itored data have been obtained and analyzed for application- oriented purposes. For example, human motions can be easily detected by using motion sensor devices such as Microsoft Kinect [1] and Leap Motion [2]. Human trajectories can be traced by using position sensing devices, such as GPS and OptiTrack [3]. These advanced sensor devices are currently used primary for analyzing object movements, as well as for visualizing them. Despite the fact that many data related to human activities, recorded by the sensors, the data exist, they cannot be managed for the purpose of processing in a unified way. In particular, motion data are stored in a proprietary format and no common formats have yet been proposed. The raw location data for human activities basically comprises a combination of human motions and positions. There are some common formats for representing human motions and positions, such as H-Anim [4] and BVH (BioVision Hierarchical data) [5]. However, these formats are used only for storage and exchange of the data, not for representations of all human activities. In the area of business process management, some process description models exist, such as BPMN (Business Process Modeling Notation) [6], XPDL (XML Process Definition Lan- guage) [7], and BPDM (Business Process Definition Meta- model) [8]. Although these models are used for business management representation and business system development, their main purpose is not to represent people’s behaviors, and thus, representations of human motions and positions are outside their scope. In this paper, we propose a new and novel process meta model for describing a model for people’s behaviors that uses a Multi-Layered Process description Model, MLPM. The model allows various kinds of human social activities, such as sport game plays, medical treatment, and agent communications, to be easily described. For example, suppose that a skilled doctor is fostering junior doctors in order to impart to them better skills for the task of giving intravenous or subcutaneous injections. It is difficult for them to understand and to perform the task without any practical experience of it. Even if they have some experience of giving the injections, appropriate real-time comments from experienced doctors are very necessary and helpful. However, when they use a subcutaneous-injection simulator, such useful comments are not available to them. There is thus no opportu- nity to improve their skills in such situations. When our model can be applied to provide this medical treatment education, a system based on our process model will be able to support young doctors by presenting to them the differences between the activity as executed by a skilled doctor and by a junior doctor, with specific details. This is because all the activities in the injection process can completely be represented by our proposed model, and the difference can be detected by real- time checking of the distance between two processes. Our proposed model is composed of seven fundamental components: Process, Task, Entity, Activity, Action, Motion, and Expression. These components are linked to many types of associations. The main characteristics of MLPM are: • All processes related to people’s behaviors can be mod- eled by using our MLPM. A process can be represented by a hierarchical structure, as well as by linked data among components. • MLPM can give researchers in many fields a way of describing behaviors in a common representation format. • Many functions and tools can be incorporated in the basic structure of MLPM. Some examples of such func- tions are process similarity searching, process outliers 167 Copyright (c) IARIA, 2014. ISBN: 978-1-61208-329-2 eKNOW 2014 : The Sixth International Conference on Information, Process, and Knowledge Management