Fast enumeration of effective mixed transports for recommending shipper collaboration
Akifumi Kira, Nobuo Terajima
TL;DR
The paper addresses fast enumeration of effective mixed transports in a logistics context, focusing on combining three lanes to minimize total loading distance. It defines the reduction rate $r$ as a distance-based metric on a metric space $(B,d)$ and develops a pruning algorithm that leverages distance bounds to prune the search for $(t_1,t_2,t_3)$ with rate $\le r$. A theoretical foundation with four lemmas ensures correctness, and a dynamic $k$-best variant using a binary heap enables efficient top-k results. Empirical results on ~17k real lanes demonstrate dramatic speedups over brute force (over $7{,}000\times$ to $>400{,}000\times$) for realistic $r$ values, highlighting the method's practical potential for cross-industry joint transportation and the TranOpt service. Overall, the work provides a scalable, exact enumeration framework that can underpin rapid suggestions of cooperative shippers in logistics networks.
Abstract
In this study, we focus on a form of joint transportation called mixed transportation and enumerate the combinations with high cooperation effects from among a number of transport lanes registered in a database (logistics big data). As a measure of the efficiency of mixed transportation, we consider the reduction rate that represents how much the total distance of loading trips is shortened by cooperation. The proposed algorithm instantly presents the set of all mixed transports with a reduction rate of a specified value or less. This algorithm is more than 7,000 times faster than simple brute force.