Prof. Won-Yong Shin from Dankook University visited our laboratory who is an ex-colleague of Prof. Ishibashi at Harvard university. He gave us an exciting and impressive talk and shared his recent results with us. The title and abstract of his talk were as follows:

Title : Huge Challenges on Twitter Analytics Using Geolocation
Abstract : Research in the field of online social networks (OSN) has grown dramatically with the evolution of technologies while harvesting Big Data. Twitter is one of the most popular micro-blogs (or social media). In this talk, I’m going to briefly explain how to perform data collection and processing from the Twitter network via Twitter Streaming API. Next, I’m going to show two recent results analyzed based on geo-tagged tweets. First, I characterize a newly discovered friendship degree according to geographic distance by introducing a new definition of “bidirectional friendship”. The study demonstrates the fact that  the number of friends according to distance follows a double power-law (i.e., double Pareto law) distribution. Second, I introduce a low-complexity algorithm that detects a “point-of-interest (POI)” boundary. Detected boundaries are shown along with a variety of POI types, and it is verified that the runtime complexity scales linearly with the input size.