Tourism News Classification Using Convolution Long Short-Term Memory (C-LSTM)

Abstract

Along with the rapid development of information technology, news about Indonesian tourism destinations can now be accessed widely through various platforms such as social media and online news, making news easily accessible. With the increase in tourism news, manual news classification is less effective in dividing data into various subcategories, such as natural tourism, artificial tourism, cultural tourism, and non-tourism. An algorithm is needed to address this problem, one of which uses an algorithm from deep learning. This study developed a tourism news classification model using Convolutional Long Short-Term Memory (C-LSTM) and Word2Vec Representation with Continuous Bag of Words (CBOW) architecture to obtain better accuracy and computational efficiency, and is used to produce better word vectorization, so that semantic relationships between words can be captured. This study used a news dataset of 5261 and news with an 80:20 ratio for training and testing. With this approach, the highest accuracy value of 94% was obtained with a time of 1140 seconds.