Translate 
Abstract
The preservation of Balinese script through digitalization faces challenges, particularly in recognizing words with multiple meanings that require different transliterations depending on context. This study developed an artificial intelligence system based on Long Short-Term Memory (LSTM) to detect the meaning of specific words in Balinese sentences and provide accurate recommendations for Balinese script transliteration. Testing showed that the LSTM model achieved an accuracy of 71% in identifying word meanings in new sentences. To enhance accuracy, this research integrated a hybrid method combining Jaro-Winkler and Damerau-Levenshtein as a preprocessing step. This combination successfully increased system accuracy to 94.58%, surpassing the previous approach, which reached 94.3% without LSTM. This integration demonstrates that the hybrid method effectively corrects spelling errors and reduces ambiguities before deep learning processing. These results represent a significant step in preserving Balinese culture through script digitalization, with further potential for development through context-based processing.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call