PeriksAksara: Web-based Indonesian Spelling and Punctuation Correction Using Jaro-Winkler Distance Algorithm Based on KBBI

Shannon Andrea, Saskia Lavinsky, Nashwa Sawitri Dewayani, Viny Christanti Mawardi · 2024

Some of the common issues that affect writing include, but not limited to - spelling errors whereby the writer may miss a certain word because of his/her negligence. This is because it is difficult for a person to observe all correct spelling rules according to the Kamus Besar Bahasa Indonesia (KBBI) also there is also a factor of speed and the movement of fingers while typing. The Jaro-winkler distance method can be used as a solution because it deals with the comparison of two small character strings which can be used to compare the given word that was misspelled with the words in the dictionary which were spelled correctly. It makes the suggestions or auto-completion more accurate and precise by giving more importance on match found at the prefix level. Experiments performed at this stage revealed that using Jaro-Winkler distance enhance the identification of spelling mistakes and generation of correct word suggestion. This result shows that the performance of this method is highly dependent on the quality of the corpus for detection and the quality of the corpus for the prediction of corrections. It means, the higher the quality of the corpus or dictionary used to generate the model, the better will be the predictions for correcting the words in the text.

Read the paper · More papers on PaperTik