Trust-Aware Personalized Route Query Using Extreme Learning Machine in Location-Based Social Networks

Abstract

The task of personalized route query is to find the optimal trip that contains the keywords specified by the query user and satisfies the travel distance constraints. The previous studies mostly focus on collaborative filtering by considering user similarity. Trust is one of the most important factors in decision-making that has been neglected by the existing studies of personalized route query. In this paper, we propose a new type of personalized route query by incorporating trust. We propose a social trust-based optimal trip selection (STOTS) framework for personalized route query. STOTS consists of three key components. The first component predicts social trust based on extreme learning machine (ELM), denoted STP-ELM, that exploits social information and user behavioral patterns as features. In the second component, we propose a novel model to incorporate social trust into personalized route query. Additionally, we propose an index to speed up query processing. In the third component, we propose an optimal route query algorithm called RouteHunter that aims to find an appropriate route satisfying the user-specified constraints. The experiment results show that (1) our social trust prediction approach based on ELM attains superior regression efficiency compared to other traditional methods; (2) our proposed index can efficiently accelerate personalized route query processing; and (3) our route query approach can achieve a better performance than the baseline approach. This paper studies a novel personalized route query incorporating social trust in location-based social networks. We propose a social trust-based optimal trip selection (STOTS) framework that uses ELM to evaluate social trust, applies a ranking model to incorporate social trust, and includes an algorithm to find the required route. Experimental results encouragingly demonstrate the efficiency and effectiveness of our proposed approach.