Length Controllable Literature Summarization Using Transformers
Vaka Avinash, Sanskriti Pattanayak, Venkatavaradan Raghuraman, Samyuktha Prakash, Arti Arya · 2022 IEEE 7th International conference for Convergence in Technology (I2CT) · 2022
This paper defines an approach to implement an output length controllable text summarizer that would help readers and learners understand large chunks of text quickly and efficiently and develop a tool to do the same. It would help and play an important role in revolutionizing today’s education. The approach defined here is a way to better leverage the current state-of-the-art (SOTA) transformer-based models to help achieve length-controllable text summaries from text corpora. The paper introduces an algorithm that, when used in unison with existing transformer-based models, enables the user to specify the required extent of summarization to generate the output summary of the desired length. The approach has been tested on a widely accepted text summarization dataset and gives acceptable ROUGE(45.31) and METEOR(22.16) scores when compared to current state-of-the-art models.