Conference Presentations by Johaira Lidasan
2024 8th International Conference on Natural Language Processing and Information Retrieval, 2024
The Maguindanaon language is one of the official languages in the Philippines that belongs to the... more The Maguindanaon language is one of the official languages in the Philippines that belongs to the Austronesian language family. However, there is a limitation of studies conducted about this language, and as a result, there is no existing Maguindanaon corpus that has been established yet. This study was primarily focused on the creation of a bilingual corpus that encompasses both Maguindanaon and English languages. The corpus was derived from everyday conversations among native Maguindanaon speakers. The primary goal was to employ a small-sized Maguindanaon corpus to develop a machine translation system. This involved using the Attention mechanism to process the data and generate translated sentences in the English language. Before evaluation, the 8,500 sentences extracted from the corpus underwent pre-processing. It was determined that the model's optimal functioning necessitates a minimum of 5,000 sentences. The model correctly assessed 38 out of 45 sentences containing known words present in the corpus. Additionally, to enhance the model's output quality, each word needs to be recurrently used in different sentences, thereby boosting its accuracy result.
4th International Conference on Computing and Machine Intelligence, 2025
This paper focuses on Maguindanaon, a language spoken by the Maguindanaon people in the Philippin... more This paper focuses on Maguindanaon, a language spoken by the Maguindanaon people in the Philippines, particularly in the provinces of Maguindanao del Norte and del Sur. With about two million speakers, the language has significant cultural value but is facing a decline as younger generations shift to more widely spoken languages. Unfortunately, there are limited resources and digital tools available to support the language, putting its future at risk. To address this, this paper develops a small-scale Maguindanaon-English bilingual corpus and uses it to test Statistical Machine Translation (SMT) models, including IBM Models 1 and 2 and a Phrase-based model. The goal was to explore the potential for using machine translation to bridge the gap between Maguindanaon and English. Although the results showed that all models struggled due to the small size of the corpus, with the Phrase-based model achieving the highest but still low BLEU scores, the study highlights the potential of machine translation for under-resourced languages. The findings emphasize the need for a larger, better-aligned corpus and improved techniques to enhance translation quality, laying the groundwork for future efforts in both language preservation and machine translation for Maguindanaon.
Papers by Johaira Lidasan
An application would be beneficial if it is real time and could<br> give its users enough i... more An application would be beneficial if it is real time and could<br> give its users enough information. This would be of greater<br> advantage for mobile applications. Mushroom Recognition using<br> Neural Network is a mobile-based application that combined the<br> power of neural network with image processing to recognize<br> mushroom image based on its order and family and if it is edible<br> or inedible/poisonous. It is a multi-class classification program<br> that recognizes mushroom image from 3 orders and 8 families<br> defined in this research. The application used the GrabCut<br> algorithm for image segmentation and Probabilistic Neural<br> Network (PNN) as its classifier that trains and classifies the<br> mushroom image. This application used 133 mushroom images<br> as its training data and obtained an accuracy rate of 92%. This<br> could be used as an educational tool both for Biology students<...
International Journal of Information Technology and Computer Science, 2019
Fever is a sign that the body is trying to fight infection. It is usually accompanied by various ... more Fever is a sign that the body is trying to fight infection. It is usually accompanied by various sicknesses or symptoms that signal another illness or disease. Diagnosing it ahead of time is essential because it has to do with human life and to determine what to do to get well. MeDevice is a mobile-based application that runs in Android devices that allows the user to enter the levels of his/her symptoms and diagnoses the disease either as influenza, dengue, chicken pox, malaria, typhoid fever, measles, Hepatitis A and pneumonia together with its details and its first aid treatment. It aims at providing an efficient decision support platform to aid people with fever in diagnosing their disease and whether or not to seek medical attention especially in developing countries like the Philippines. This application is engineered with the knowledge base and the inference method of fuzzy logic and expert system with the help of Gradient Descent optimization algorithm and back propagation neural network to achieve the optimum value of the error rate. This is essential to provide the application with a high accuracy rate which shows during the conduct of testing of the application.
Uploads
Conference Presentations by Johaira Lidasan
Papers by Johaira Lidasan