Real-Time EEG-Based Emotion-Adaptive Brain–Computer Interface for Intelligent NPC Interaction in Virtual Environments

The dataset accompanying this manuscript contains all quantitative results used to generate the figures and tables presented in the paper “Real-Time EEG-Based Emotion-Adaptive Brain–Computer Interface for Intelligent NPC Interaction in Virtual Environments.” The data are organised into one Excel file and multiple CSV files to ensure transparency, reproducibility, and ease of reuse. The file BAC_All_Tables.xlsx contains all tabulated results used in the manuscript, with each table stored in a separate worksheet: Table 1: Feature vector composition for EEG-based emotion recognition. Table 2: Experimental setup parameters, including participants, hardware, channels, and session details. Table 3: Per-class precision, recall, and F1-scores for real-time emotion classification. Table 4: End-to-end latency breakdown across pipeline stages. Table 5: Mean user study ratings for responsiveness, empathy, realism, and engagement. The figure data are provided as individual CSV files: Figure5_ConfusionMatrix.csv – Confusion matrix values for five emotion classes. Figure6_EmotionTimeline.csv – Time-series data representing dominant emotional states over a 200 s session. Figure7_LatencyHistogram.csv – Sampled end-to-end latency values for histogram and trend analysis. Figure8_UserRatings.csv – Mean user ratings across interaction quality dimensions. Figure9_Ablation.csv – Engagement scores for static, MLP-only, and hybrid NPC configurations. All files are in open, non-proprietary formats and can be directly used to reproduce the figures and tables in the manuscript or to support further analysis and comparative studies in affective brain–computer interface research.