Implementasi Natural Language Understanding pada Sistem Chatbot Tahapan Audit K3 menggunakan Algoritma Long Short-Term Memory (Lstm)

Abstract

Health, Safety and Enviroment (HSE) has known become more prominent in companies with the primary goal of protecting all workers from accidents and minimizing losses in both property and financial aspects caused in their workplace. To reduce all unwanted incidents, strict procedures and audits periodically are necessary. Audits typically require more attention to detail and complex procedures to ensure all work to do done properly. all findings must be followed up and trends identified across all categories, types, elements, and statuses of the findings. Given the sheer number of factors and data involved, manually processing and validating this information by the internal HSE audit team to ensure that partners have followed the standard procedures and regulations can be very time-consuming. To maximize the efficiency this process, a technological approach need to be involve like chatbot. Using chatbot makes the validation process much more efficient and quicker, enabling faster determination of audit findings. In this study, the approach uses Long Short-Term Memory (LSTM) and Natural Language Understanding (NLU) to recognize the intent behind each question asked. The research utilizes data from questions submitted by work partners, categorizing each question based on its respective intent so that the chatbot can provide answers in line with the internal audit’s guidelines. This study result show that the chatbot achieved an accuracy rate of 84% and user satisfaction 80%.