Simplified animation circuit for metadata-based behavior model query and retrieval

This paper describes an approach to simplify the animation circuit so that useful metadata could be extracted to facilitate behavior model storage and retrieval. In the VR model of a product, the tasks performed by the behavior nodes could be classified into three categories, i.e., changing the attributes of a part, animating the behavior of a part, and detecting user's interaction with a part. Based on this property, 2-tuple metadata (P, T) is designed to represent the behavior node, where P is the part operated by behavior node and T is tasks performed by the behavior node. The parts and tasks contained in the metadata are hierarchical classified. The process to simplify the animation circuit consists of two major steps. First, the behavior nodes in the animation circuit are converted to their metadata. The resulted graph is called a metadata graph. Secondly, the metadata graph is simplified by merging nodes that have the same classifications. The structured metadata extracted from the metadata graphs allows behavior model query and retrieval to be performed at different levels of abstractions.