A PROPOSED INTEGRATED FRAMEWORK FOR IDENTIFYING SPECIAL SIGNAL PATTERNS FROM PSEUDO-REAL TIME DATA: APPLICATION TO GEOCODED SOCIAL MEDIA DATA

摘要

The Internet and information communication technologies (ICT) are becoming an integral part of our daily lives. Now, with a simple touch of our smartphones, information on most any topic can be accessed and shared in matter of seconds. The manner information flows today has revolutionized the way we communicate and receive information with regards to world events and disasters. Social media has become an important source of real time information as more people are connected online and are able to share content (e.g., texts, pictures, videos, etc.) related to events that are of importance to them. In particular, disasters present a scenario where social behavior dictates a reaction that is different from routine situations. More and more, social media has become an effective alternative to traditional media in disseminating information on current events and the status of the environment around those events. In this paper, we lay down the theory to develop a methodology to integrate signal detection theory and statistical process control into a single framework for identifying patterns/signals in pseudo-real time data feeds. We hypothesize that signals (e.g., from geocoded social media generated data) follow patterns that can be linked to a specific cause or event. The theoretical framework addresses the challenge of integrating signal detection theory (SDT) and statistical process control (SPC) in a single and simple framework for identifying specific event patterns/signals from near-real time big data.