Avatar Path Clustering for Networked Virtual Environments

• Main Authors: Ching-Chuan Huang, 黃景詮
• Other Authors: Jehn-Ruey Jiang
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/52956798332551343006

碩士 === 國立中央大學 === 資訊工程研究所 === 98 === With the increase of network bandwidth and the advance of 3D graphics technology, networked virtual environments (NVEs) have become one of the most popular research topics. Massively multiplayer online games (MMOGs), like Second Life and World of Warcraft (WoW), are well known examples of NVEs. Because users’ interests or habits may be similar, avatars, the representative of users on the NVE, may have similar behavior patterns, which leads to similar paths on the NVE. For instance, different kinds of virtual shops in Second Life attract a variety of users to drop by, and thus users with similar interests usually head toward common destinations and produce similar paths. This research proposes two avatar path clustering algorithms for NVEs, namely, Average Distance of Corresponding Points-Density Clustering (ADOCP-DC) and Longest Common Subsequence-Density Clustering (LCSS-DC). In ADOCP-DC algorithm, the path similarities are computed with the ADOCP mechanism first, and then paths are clustered with Density-Based Clustering to find users with similar paths. LCSS-DC algorithm uses the LCSS mechanism to compute the path similarities and then clusters paths with Density-Based Clustering. Both algorithms will produce a Representative Path (RP) for each cluster of paths. They can be applied to several research areas like peer-to-peer networked virtual environments (P2P-NVEs), avatar state management, avatar behavior analysis and server load balancing. Game developer can also apply the algorithms to find out popular paths for improving the game design. We take Second Life user trace data as input of the algorithms to demonstrate the algorithms’ execution. We also show how to adjust algorithm parameters to obtain high-quality path clustering.