Bengali News Headline Generation on the Basis of Sequence to Sequence Learning Using Bi-Directional RNN

Abstract

The newspaper provides significant information every day. Each news describes every event largely, but the headline contained a summary of the news. A meaningful headline is very good for a reader who can understand the gist of corresponding news. Text generation is a language modelling where the machine can generate text automatically. Predict the next correct sequence of a text in the main concept of text generation. Using the text generation approach, automatic headline generation is the possible solution for any language. Many experiments have already been completed for English text generation but few in the Bengali language. Here, it has a big scope to generating an automatic news headlines generator for Bengali text using deep learning concept with bi-directional recurrent neural network (RNN). Bengali text summarization is our recent research work in Bengali NLP. During this research, Bengali text generation for text summarizer is challenging work. Thus, to solve these types of problem, automatic text generation is a good solution.