Hand gesture recognition has many important applications in Human-Computer Interactions (HCI). Various approaches and methods have been used by researchers to recognize gestural motions. In addition, the application of human motion interaction has been applied on modern technologies such as wearable wireless devices, computer game and smartphone companion gadget. Previous researcher mostly focusses on variation of sensor type to record arm movement and different algorithms to process the input gesture. Various features of arm movements are extracted in order to increase the trait of the intelligent systems. However, the result of the recognition accuracy is still can be improved by avoiding data losses on important characteristic features of arm movements. Beside, some important characteristic features will loss during features extraction process. Thus, the concept of ontology for describing the architecture of arm movements is seen possible to overcome the sustainability and accuracy problems of the intelligent systems. This paper discusses about tracking and designing pattern trajectories of the arm movement model, based on concepts of knowledge-based ontology. Knowledge is the initial stage from three domains in ontology framework. It is the phase where all the input data gathered and analyzed before matching and recognition process are conducted in process domain. In this paper, 2D raw motion data of arm gesture presented by x and y-axes coordinate from a motion capture system (MOCAP), were sampled down and converted into distance before it's were normalized and known as knowledge data. This paper shows the knowledge data is essential to build the attribute domain. Attribute domain is the second domain in the ontology framework, consists of the derived element from knowledge data, which is speed and acceleration.
