Word Segmentation Task for Southeast Asian Abugida Scripts: A Systematic Literature Review

Word segmentation is a part of natural language processing and computational linguistics that is used to find word boundaries in a text. Many Southeast Asian Abugida scripts, including Khmer, Lao, Thai, Burmese, Javanese, and Sundanese, commonly employ scriptio continua, a writing style in which no spaces or other markings are used between words or sentences. The scriptio continua writing style presents a distinct ambiguity problem for the word segmentation process. This review aims to provide a thorough overview of the word segmentation task for Southeast Asian Abugida scripts that use the scriptio continua writing style. Understanding word segmentation tasks is crucial for NLP research and applications, such as text analysis, machine translation, and information retrieval. Word segmentation accuracy is commonly measured using the F1 score. This review is organized into chapters that describe the research methodology, results, and analysis before closing with a summary of the findings. This review found that three unique scripts—Thai, Burmese, and Khmer—have been thoroughly researched over the past five years. Combining different approaches, such as traditional-based and machine learning, has been employed to maximize the accuracy of word segmentation tasks for these scripts. The most common data sources processed for these tasks are derived from established corpora.