Our laboratory published an article in the international journal Big Data and Cognitive Computing (Scopus, Q1):

Ryumina E., Ryumin D., Markitantov M., Karpov A. ExPAM: Explainable Personality Assessment Method Using Heterogeneous Linguistic Features and Off-the-Shelf LLMs // Big Data and Cognitive Computing, 2026, vol. 10(8), 254. (WOS IF=5.3 Q1, Scopus SJR=1.033 Q1)

Automatic personality assessment plays an important role in personalized intelligent systems; however, existing approaches do not always provide both high-quality predictions and clear interpretability. The paper presents a novel explainable personality assessment method, ExPAM (Explainable Personality Assessment Method), designed to predict Big Five personality traits from textual data. ExPAM combines deep contextual features obtained using Transformer-based models with interpretable linguistic characteristics extracted using the Linguistic Inquiry and Word Count (LIWC) dictionary. The resulting global and local linguistic patterns are incorporated into prompts for off-the-shelf Large Language Models (LLMs), enabling personality-trait predictions to be refined while simultaneously generating human-understandable explanations without additional LLM fine-tuning. Experiments are conducted on the ChaLearn First Impressions v2 (FIv2) and PANDORA corpora. On FIv2, the method achieves a mean accuracy of mAC=0.891 and a Concordance Correlation Coefficient of CCC=0.333, while on PANDORA it achieves a mean Pearson Correlation Coefficient of PCC=0.240 and CCC=0.101. Incorporating hybrid global–local linguistic patterns into LLM prompts further improves CCC by 9.9% on FIv2 and by 15.8% on PANDORA. Interpretability analysis reveals linguistic patterns associated with individual personality traits, demonstrating the potential of ExPAM for research in psychology, computational linguistics, and paralinguistics.