Discovering Stable Ride-Sharing Groups for Commuting Private Car Using Spatio-Temporal Semantic Similarity

Yuhui Ye, Linjiang Zheng, Yiling Chen, Longquan Liao

Abstract: The commuting behavior of private cars is the main cause of traffic congestion in the morning and evening rush hours. The discovery of stable ride-sharing groups for commuting private car is a necessary work to solve the morning and evening peak traffic congestion, and the discovery of ride-sharing groups is to match similar travelers. But different from the traditional ride sharing service based on passenger request, commuting private car users expect a long-term stable ride sharing service. This paper proposed a commuting private car ride-sharing groups discovery model based on spatio-temporal semantic similarity. The model considers the similarity of commuting time and space, and adds the user feature of commuting workplace’s semantics to provide stable sharing matching for commuters of similar occupations, and adds the semantic of commuting place to provide stable ride-sharing matching for commuters with similar occupations. Commuting workplace’s semantics refers to the functional area type of commuting workplace, which can reflect the nature of commuters' work. The main research content includes two parts: functional area identification and ride-sharing groups’ discovery. In the process of functional area recognition, based on POI and road network data, multi classification support vector machine is used to identify urban functional areas. In the process of ride-sharing groups’ discovery, based on the travel data of commuter private cars, DBSCAN clustering algorithm with spatio-temporal semantic distance measurement is used to complete ride-sharing group discovery.