The study aims to develop an application that, conditioned on an image, generates a Filipino news article that is similar to a real article. It explores the capabilities of current NLP algorithms for language generation in the low-resource Filipino language. In order to carry out the study, a dataset of Filipino news articles with accompanying images will be collected. With this, the study also sets out to publish a dataset for conditional language generation in the Filipino language.
Transformer-based Conditional Language Models to Generate Filipino News Articles
51 Views
7 Downloads