Designing a Chatbot Using Long Short Term Memory Algorithm to Evaluate the Accuracy in Comparison with Naive Bayes Algorithm

The objective of this study is to assess the accuracy of a chatbot utilizing the LSTM (LSTM) algorithm, comparing it to the Naive Bayes algorithm. The evaluation focuses on determining which algorithm delivers more accurate responses to user queries. The research involves two distinct groups: Group 1, which employs the Novel LSTM algorithm, and Group 2, which uses the Naive Bayes algorithm. Each group consists of 20 samples, determined based on a pretest power of 80 % and an error rate of 0.05, resulting in a total sample size of 40. The findings reveal that the chatbot powered by the Novel LSTM algorithm achieves an accuracy of 95.66 %, surpassing the 94.6 % accuracy of the Naive Bayes algorithm. The statistical analysis indicates a significance level of $0.011(p<0.05)$, highlighting that the LSTM-based chatbot significantly outperforms the Naive Bayes-based chatbot in terms of accuracy. This result underscores the superior performance of the Novel LSTM algorithm in providing precise and reliable responses compared to its Naive Bayes counterpart.