INTEGRASI ALGORITMA NLP UNTUK PENINGKATAN KECERDASAN CHATBOT: KAJIAN LITERATUR DAN ANALISIS PERKEMBANGAN TERKINI

Research on Natural Language Processing (NLP) algorithms in chatbot development has increased significantly, driven by the demand for intelligent human-like conversational interactions. This study aims to map NLP algorithm trends, application domains, and future research directions. The method employed is a systematic literature review of 103 publications published between 2020 and 2025, involving systematic selection processes from duplicate filtering to in-depth evaluation of research methods and results. The analysis shows a dominance of Transformer architectures, such as BERT, ELECTRA, and IndoBERT. Nonetheless, traditional algorithms like LSTM, Naive Bayes, and SVM are still utilized for lightweight computation or comparative purposes. In terms of application domains, most research is concentrated in General, Health, and Education sectors, while Law, Banking, and Human Resources remain under-explored. The findings also indicate a gap between technical development and production-scale implementation, along with a lack of focus on ethics and data protection. In conclusion, future research should focus on chatbot personalization, exploration of new domains, integration of ethical aspects, and the implementation of explainable AI (XAI) to enhance transparency and public trust.