Backward Path Growth for Efficient Mobile Sequential Recommendation

TL;DR

This paper introduces a dynamic programming approach with backward path growth and pruning techniques to efficiently solve the mobile sequential recommendation problem, optimizing taxi routes for passenger pickup with reduced computational cost.

Contribution

It proposes a novel backward incremental sequence generation algorithm with pruning policies, significantly improving efficiency over existing methods for large-scale route optimization.

Findings

Pruning percentage is significantly improved compared to state-of-the-art methods.

The method effectively handles more pick-up points and arbitrary route lengths.

Experimental results demonstrate enhanced efficiency and scalability.

Abstract

The problem of mobile sequential recommendation is presented to suggest a route connecting some pick-up points for a taxi driver so that he/she is more likely to get passengers with less travel cost. Essentially, a key challenge of this problem is its high computational complexity. In this paper, we propose a dynamical programming based method to solve this problem. Our method consists of two separate stages: an offline pre-processing stage and an online search stage. The offline stage pre-computes optimal potential sequence candidates from a set of pick-up points, and the online stage selects the optimal driving route based on the pre-computed sequences with the current position of an empty taxi. Specifically, for the offline pre-computation, a backward incremental sequence generation algorithm is proposed based on the iterative property of the cost function. Simultaneously, an incremental pruning policy is adopted in the process of sequence generation to reduce the search space of the potential sequences effectively. In addition, a batch pruning algorithm can also be applied to the generated potential sequences to remove the non-optimal ones of a certain length. Since the pruning effect continuously increases with the increase of the sequence length, our method can search the optimal driving route efficiently in the remaining potential sequence candidates. Experimental results on real and synthetic data sets show that the pruning percentage of our method is significantly improved compared to the state-of-the-art methods, which makes our method can be used to handle the problem of mobile sequential recommendation with more pick-up points and to search the optimal driving routes in arbitrary length ranges.