A Comparative Analysis of Counselor and ChatGPT Texts in Adolescent Counseling Using Text Mining

This study uses text mining techniques to compare adolescent counseling texts provided by human counselors in an online forum counseling setting and to offer implications in AI-based counseling. A total of 900 responses were analyzed, comprising 450 cases each from an online counseling platform and ChatGPT-generated responses based on the same client cases. Word frequency, semantic network, and sentiment analysis, were conducted. The results are as follows: First, the word frequency analysis revealed that counselors used twice as many keywords and emphasized social and interpersonal contexts, while ChatGPT focused on individual problem-solving. Second, the semantic network analysis showed that counselors facilitate clients’ self-reflection, whereas ChatGPT highlighted personal values and emotions. Third, the sentiment analysis indicated that counselors used a wider range of emotional expressions. Implications and future suggestions for AI-based adolescent counseling were discussed.