One of the challenges of emotion classification is the existence of low annotated datasets, that makes the task more complex. Certain existing datasets often suffer from imbalanced data for the emotion classes. Several data augmentation approaches can help to overcome the challenges regarding imbalanced datasets. However, the existing data augmentation techniques in emotion classification lack consideration for the contextual nuances of emotions and this area is still relatively underexplored. In this work, we study the impact of data augmentation on classification performance of three machine learning models including Logistic Regression, BiLSTM and BERT and compare frequently used methods to address the issue. Specifically, we assessed Easy Data Augmentation (EDA) and contextual Embedding-based data augmentation (BERT) on two datasets. Based on the experimental results, we combined two BERT-based augmentation techniques including insert and substitute, to generate data for minority emotion classes. Furthermore, we proposed a data augmentation method using ChatGPT. Compared to the baseline models, incorporating the BERT augmentation techniques with BERT model resulted in improvements of +4.34% and +5.56% in Macro F1 score on the SemEval-2018 and GoEmotions datasets, respectively. Moreover, the proposed augmentation technique utilizing ChatGPT yielded improvements of +3.55% and +4.83% on the same datasets.

The Vehicle Routing Problem with Time Windows (VRPTW) is a critical combinatorial optimization problem in modern logistics, where finding optimal routes is essential for minimizing operational costs and enhancing service reliability. While Genetic Algorithms (GAs) are a powerful tool for solving VRPTW, their effectiveness is often undermined by premature convergence, a phenomenon in which the algorithm stagnates at suboptimal solutions, thus failing to achieve maximum efficiency. This study directly addresses this challenge by systematically evaluating how different reproduction schemes impact GA performance. The primary objective is to identify operator combinations that mitigate premature convergence to achieve superior solution quality, measured by total travel time, while also analyzing the trade-off with computational cost. We investigate combinations of conventional operators, such as Tournament Selection (TS) and Order Crossover (OX), against more advanced schemes, including Split Rank Selection (SRS) and Multi-Parent Order Crossover (MPOX), as well as different mutation methods, namely Scramble Mutation (SM) and Inversion Mutation (IM). Results demonstrate that advanced schemes, particularly the combination of SRS, MPOX, and IM, yield the most robust convergence and the lowest average total travel time of 48,422.8 minutes. However, this superior performance requires the longest computation time at 30.9 h. In contrast, conventional operator combinations are significantly faster, with execution times as low as 8.7 h, but they produce lower-quality solutions and exhibit unstable convergence. This study highlights the crucial role of the reproduction scheme in balancing the trade-off between solution quality and computational efficiency, confirming that a synergistic combination of advanced operators is essential for solving complex VRPTW instances effectively.