Method for Combining Data Farming and Data Mining in a Logistics Assistance System for Materials Trading Networks Based on Graph Databases

To maintain the competitiveness of a materials trading network, decision-makers are confronted with a multitude of logistics tasks. Finding answers to these tasks often involves a decision-making process, which in turn requires a detailed analysis and evaluation of the state of the materials trading network. Typically, logistics assistance systems are used for this purpose, as they include various methods for this purpose, such as simulation. This dissertation develops a novel method for logistics assistance systems by combining simulation-based data generation, called data farming, and knowledge discovery in the domain of materials trading networks. By combining data farming and knowledge discovery, logistics tasks can be addressed in a targeted manner and the knowledge gained can be made available to the decision-makers of a materials trading company. The method includes a modeling concept for developing a simulation model using labeled property graphs, integrates data storage in graph databases, and motivates the use of mining algorithms suitable for graph data. The method is evaluated, and its applicability is demonstrated via a use case based on observational data from a materials trading company. A critical re ection illustrates the feasibility of the method, highlights advantages, and discusses limitations