Indonesian Lexical Ambiguity in Machine Translation: A Literature Review

Machine Translation is essential for transforming text from one language, such as Indonesian, to another language while maintaining the original meaning. However, this task faces challenges related to lexical ambiguity, same word but has very different meanings. This paper offers an overview of recent research on ambiguity handling in Indonesian Machine Translation, focusing on studies published between 2014 and 2024. The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) and PICO (Population, Intervention, Comparison, Outcome) framework was used to conduct the systematic literature review of Indonesian Machine Translation to address the challenge of lexical ambiguity. By 27 primary studies we found, the majority rarely discuss the handling of ambiguity, especially lexical ambiguity. However, several key themes emerge among the studies that address ambiguity. These include issues such as homonyms, polysemy, informal language, and disfluency. Various methods are explored to tackle these ambiguity problems, including partof-speech (POS) tagging, word feature extraction, and semantic similarity measures. In conclusion, while ambiguity handling remains an under-explored aspect in the development of Indonesian Machine Translation, the reviewed studies emphasize the importance of addressing this challenge and offer promising methodologies for future research in this area.