On Application of Learning to Rank for Assets Management: Warehouses Ranking

Abstract

The number of connected Internet of Things devices is forecast to grow to more than 31 billion globally in 2018. Many of these devices are/will be connecting to physical assets, and data collected from these devices can be used to model, manage, or describe these assets. We are motivated by utilizing these abundant data for assets’ resources requirements ranking. Multiple Criteria Decision Analysis, MCDA, has often been utilized in finding ranking by computing on collected data and predefined criteria. However, changes in asset’s environment have a direct impact on its resource requirements ranking and decisions on ranking must be constantly revised. This is a repetitive process where managers require to repetitively adjusting the decisions on the ranking. With machine learning, such repetitiveness can be heavily minimized by teaching machines how to rank not by instruction but rather by examples of the task being done. In this paper, we present Learn to Rank, LTR, machine learning framework in conjunction with MCDA for warehouse resources requirements ranking application. A framework for smart contract renegotiate resources allocation ranking for each asset on the blockchains is also discussed.