Research On Unified University Social Network Platform And Campus Social Behavioral Data Collection Mechanisms

As the Chinese economy processes through reform and construction and emerging indus-tries undergoes rapid growth, the Internet industry in particular is transforming social life at a pace faster than ever. However the IT infrastructure renewal effort of Chinese universities, in contrast, are comparatively lackluster and increasingly out of sync with contemporary commer?cial offerings. Chinese universities are set up for training professionals for construction of the country; researches and improvements to the social relationships of their students and faculties thus have profound implications, in that they enable optimization of both teaching procedure and result, but the current literature are scattered and non-systematic, in-depth but lacking in integration. Thus it is considered necessary to build a research platform in an effort to spear-head academic interest, to bind together the interests of universities and industries, and to enable constructive reformation of campus social life and information flows.Focusing on entities involved in the Jiangnan University campus, use cases of the different entities are analyzed. Extending the observation to generic universities and other organizations,an entity relationship between individuals and organizations is established. Upon this a uni-fied university social network platform is architected. By segregating organization-specific and organization-agnostic logic, a system topology laid upon the Docker containerization infras-tructure is proposed. An Android client is also designed with a plug-in architecture.Research of social relationships rely heavily on collection of behavioral data and meta-data. As an infrastructure-level research, a strategy for collection of university social interac-tion data is formed at a framework level. Behavioral data is collected at both the API backend and client-side. Behavioral metadata consists of device fingerprints and location information.Based on the Android smartphone platform, a scheme for deriving device fingerprint is devel-oped. And a WLAN-based in-door location system is developed with the observation of the partially controlled nature of campus wireless access points. Experiments done on Jiangnan University campus show good accuracy for use in university in-door environments.Aside from the campus community, the live-streaming industry is also receiving signifi-cant attention in recent years. Campus live-streaming is emerging as a new means of socializ-ing. Live streams are typically accompanied with real-time comments known as danmaku, of which proper supervision and analysis is necessary whether external or on-campus. To serve as pre-research for future campus live-streaming danmaku services, an algorithm for new word discovery in Chinese danmaku is devised based on existing mathematical constructions of in-ternal cohesion and boundary freedom measures. Experiments done on a danmaku corpus from a well-known commercial live-streaming service show promising run-time performance and word extraction ability, and adaptability to differently sized inputs.