Matrix computations (3rd ed.)
Virtual Institutions: Normative Environments Facilitating Imitation Learning in Virtual Agents
IVA '08 Proceedings of the 8th international conference on Intelligent Virtual Agents
Intelligent Agents Living in Social Virtual Environments --- Bringing Max into Second Life
IVA '08 Proceedings of the 8th international conference on Intelligent Virtual Agents
Relational learning via latent social dimensions
Proceedings of the 15th ACM SIGKDD international conference on Knowledge discovery and data mining
Proceedings of the 1st Conference of the Extreme Science and Engineering Discovery Environment: Bridging from the eXtreme to the campus and beyond
Hi-index | 0.00 |
Virtual worlds and massively multi-player online games are rich sources of information about large-scale teams and groups, offering the tantalizing possibility of harvesting data about group formation, social networks, and network evolution. However these environments lack many of the cues that facilitate natural language processing in other conversational settings and different types of social media. Public chat data often features players who speak simultaneously, use jargon and emoticons, and only erratically adhere to conversational norms. In this paper, we present techniques for inferring the existence of social links from unstructured conversational data collected from groups of participants in the Second Life virtual world. We present an algorithm for addressing this problem, Shallow Semantic Temporal Overlap (SSTO), that combines temporal and language information to create directional links between participants, and a second approach that relies on temporal overlap alone to create undirected links between participants. Relying on temporal overlap is noisy, resulting in a low precision and networks with many extraneous links. In this paper, we demonstrate that we can ameliorate this problem by using network modularity optimization to perform community detection in the noisy networks and severing cross-community links. Although using the content of the communications still results in the best performance, community detection is effective as a noise reduction technique for eliminating the extra links created by temporal overlap alone.