NAMED ENTITY RECOGNITION PADA TEKS BERBAHASA INDONESIA MENGGUNAKAN CNNs

Abstract

Abstract
This research aims to develop and evaluate a Named Entity Recognition (NER) model on Indonesian text using
Convolutional Neural Networks (CNNs) architecture. The research process begins with the collection of datasets
involving news articles, formal documents, and voice-to-text transcriptions, using web scraping techniques to
obtain data automatically. This dataset then goes through a preprocessing process that includes case folding and
tokenization. Next, the data is labeled using the Stanford BIO format to identify entities. Data division was done
into training, testing, and validation sets with various scenarios. Models were trained and evaluated based on
precision, recall, and F1-score metrics to measure performance in identifying entities such as names of people,
organizations, and locations. The results show that the CNNs approach is effective in improving entity recognition
accuracy in Indonesian text, with a high F1 Score for different entity categories, indicating significant potential
for further applications in natural language processing in Indonesia