Abstract

Javanese language can be said to be a unique language because Javanese language has many meanings even though one word is the same but different regions. The Javanese are the largest tribe of 41% or about 95,217,022 people. Java island is also the largest internet access, especially for social media, which is 87.13%, from the data a lot of information can be obtained from the activeness of Javanese people in using social media. Especially textual data, but it is not easy to get javanese expressions or emotions from social media because the amount of data is very large. Therefore, this problem can be analyzed. Text mining is one of the right ways because social media data is more textual data. Using RNN Architecture and Longs Short Term Memory (LSTM) algorithm, textual data will be easy to process because a lot of data will be spruced up and calcified. So that a lot of data will be selected and processed according to the expression, because this study focuses on expression and emotions then the data will be classified with LSTM. From the results of the process of using LSTM to sort expressions into 4 namely angry, happy, sad, and afraid to get 92% accuracy. This accuracy indicates that Long Short Term Memory (LSTM) is effective in classifying Javanese text expressions.