Personalized Visited-POI Assignment to Individual Raw GPS Trajectories

Abstract

Knowledge discovery from GPS trajectory data is an essential topic in several scientific areas, including data mining, human behavior analysis, and user modeling. This article proposes a task that assigns personalized visited points of interest (POIs). Its goal is to assign every fine-grain location (i.e., POIs) that a user actually visited, which we call visited-POI , to the corresponding span of his or her (personal) GPS trajectories. We also introduce a novel algorithm to solve this assignment task. First, we exhaustively extract stay-points as span candidates of visits using a variant of a conventional stay-point extraction method and then extract POIs that are located close to the extracted stay-points as visited-POI candidates. Then, we simultaneously predict which stay-points and POIs can be actual user visits by considering various aspects, which we formulate as integer linear programming. Experimental results conducted on a real user dataset show that our method achieves higher accuracy in the visited-POI assignment task than the various cascaded procedures of conventional methods.