A BSC-based network DEA model equipped with computational linguistics for performance assessment and improvement

This research introduces a fusion architecture that integrates balanced scorecards (BSCs) and network data envelopment analysis (NDEA) to conduct a performance evaluation task from multiple perspectives. The architecture is able to capture the dynamics of production processes and sub-processes, uncover some of the components behind successful business practices, and shed light on needed actions for decision makers. Furthermore, the architecture not only can support decision makers to plan for improvement, but also equip them with forecasting ability. To enhance its forecasting quality, this study goes beyond quantitative ratios and extends them to qualitative ratios (i.e., readability: the complexities of disclosure) borrowed from computational linguistics. The results indicate that a poor readability score is highly associated with bad operations. Finally, to enlarge the mechanism’s applicable fields, the study executes the genetic algorithm (GA) to extract the inherent decision logics and represents them in a human-readable manner. The mechanism, examined by real cases, is a promising alternative for performance evaluation and forecasting.