Implementasi Long Short-Term Memory (LSTM) dan Integer Sequence Matching pada Sistem Chatbot Informasi Saham Indonesia

Abstract

The percentage of Indonesians investing in the Indonesian stock market is still relatively low compared to other countries. This is due to a lack of information about stocks among the Indonesian population. This problem is a key reason for the need for accessible stock information tools, one of which is a chatbot. Besides being accessible anytime and anywhere, a chatbot also serves as a two-way question-and-answer platform, making the distribution of information much easier. The development of a stock chatbot using Long Short-Term Memory (LSTM) methods and Integer Sequence Matching can help address this issue by providing answers to user queries about stocks. It also includes supporting features such as a stock learning module, information on stock sectors, information on stock types per sector, and real-time stock prices, all of which can attract the interest of Indonesian investors. The data used is obtained from OJK-certified securities to ensure high-quality answers. Evaluation shows that the resulting chatbot has an accuracy rate of 90% with an average response time of around 0.89 seconds. The chatbot is positively assessed by the general public, including 30 persons who have never invested in Indonesian stocks, 30 persons who have invested in Indonesian stocks, and 4 Indonesian stock experts. It is considered to be easy to understand and bring benefit. It is assessed to enhance interest of 24 out of 30 persons who have never invested in Indonesian stocks to invest stock in Indonesia stock market.