Implementation of Pattern-Based Natural Language Processing (NLP) Method on Chatbot

Abstract

Natural Language Processing (NLP) based chatbots are increasingly used in various applications, ranging from customer service to education. One simple yet effective approach to implementing NLP on a chatbot is the pattern-based method, which uses language patterns to understand and respond to user input. This research aims to implement the pattern-based NLP method on a chatbot and evaluate its performance using the BLEU metric. Conversation data is collected and analyzed to identify common patterns, which are then mapped into appropriate responses. Testing was done by measuring the level of bigram match between the chatbot response and the reference response, where the BLEU-2 precision reached a value of 0.75 or 75%. These results show that the pattern-based method is capable of generating relevant responses but still requires refinement to achieve higher accuracy.