Malagasy Abstractive Text Summarization Using Scheduled Sampling Model
Volatiana Marielle Ratianantitra, Jean Luc Razafindramintsa, Thomas Mahatody, Claire Rasoamalalavao, Victor Manantsoa · Advances in intelligent systems research/Advances in Intelligent Systems Research · 2022
Since 1955, text summarizing has evolved.We could observe all the various approaches in several languages, although most of these methods were for significant languages such as English, French, etc.Other scholars have figured out how to summarize their material in their language (a language other than the major languages), which has led us to discover a way for our language, the Malagasy language, which is considered an under-endowed language.An abstractive text summarizing approach is presented in this study.The abstractive technique is more complicated than the extractive approach because it entails re-formulating the source material while maintaining the general idea.However, it results in a more natural summary and better sentence harmony.The Scheduled Sampling approach was utilized to develop the text summarization model, which used deep learning.The task at hand is to teach the model how to communicate in English.The obtained results suggest that deep learning may be applied to the Malagasy language.