Route Recommendation Method Based on Frequent Trajectory Sequence Pattern Mining

Abstract

Travel route recommendation is one of the important research contents in the field of intelligent transportation. Traditional route recommendation methods often recommend routes on the basis of a single factor such as the shortest route or the shortest travel time, while ignoring the influence of urban crowd travel patterns on the route recommendation process. For the problems, in this paper a route recommendation method based on frequent trajectory sequence patterns was proposed. In the data preprocessing stage, the historical trajectory database is used to mine frequent sequence patterns in different periods of city and build a frequent route sequence pattern database. In the path recommendation stage, for a set of candidate paths determined after a given start and end point, the proposed long-short mode weight evaluation model is used to quantitatively evaluate and rank them. Then, the path of which the evaluation value is Top-n is taken out and recommended for the user. The recommendation results were analyzed through four simulation scenarios, and the results show that the recommendation method is reasonable. Compared with the shortest path and test set, the recommended path is better, and it runs faster than traditional path recommendation algorithm.