Generative Language Model for Chatbots Based on TF‐IDF Personality Feature Recognition and Classification

With the increasing demand for intelligent human–computer interaction, to enhance the personalized expression ability and contextual adaptability of language generation for chatbots, a user‐group personality feature recognition model based on the term frequency‐inverse document frequency model is designed. By analyzing the frequency and semantic distribution of keywords in user historical texts, tag classification of five personality dimensions is achieved. Subsequently, the classification results are embedded as control signals into an improved sequence structure language generation framework, introducing attention enhancement mechanisms and personality guidance vectors to enhance the model′s ability to control tone style and contextual logic. The experimental results showed that the proposed model achieved the highest F1 score of 93.65% in language generation on two public datasets, with an average processing delay controlled within 210 ms, which was significantly better than other mainstream models. The Top‐1 hit rate remained above 90% in both anger and sorrow contexts, with an average user satisfaction rating of 4.48 points. The proposed model has significant advantages in improving the personalization, contextual perception, and user interaction satisfaction of chatbot language generation, providing valuable technical support for the application of intelligent dialogue systems in educational counseling, psychological companionship, and service robots.