1. Development of detection and correction of errors in spelling and compound words using long short-term memory.
- Author
-
Dwitya, Nabil Rakha and Overbeek, Marlinda Vasty
- Subjects
- *
COMPOUND words , *SPELLING errors , *DEEP learning , *INFORMATION sharing , *ORTHOGRAPHY & spelling , *NATURAL language processing - Abstract
Language serves as a crucial tool for human communication, facilitating the exchange of information through various mediums, including journalism. In Indonesia, adherence to proper language rules, including word writing conventions, is essential. However, despite advancements in technology, errors in grammar and spelling remain prevalent in online news portals, leading to ambiguity in conveyed information. This research addresses the correction of both spelling and compound word errors using Natural Language Processing (NLP) technology, particularly employing the Long Short-Term Memory (LSTM) Deep Learning approach. The models developed demonstrate significant accuracy, with the compound word error detection model achieving 98.14% accuracy and the spelling error model reaching 87.12%. Although the primary focus is on compound word error detection, the supplementary detection and correction of spelling errors aim to enhance the overall accuracy of compound word detection in the inputted articles. Testing conducted on real-case articles further validates the effectiveness of the models, with an average accuracy, precision, recall, and F1-Score rate of approximately 96.45%, 93.92%, 100%, and 96.85%, respectively, across different sets of articles. [ABSTRACT FROM AUTHOR]
- Published
- 2024
- Full Text
- View/download PDF