Leveraging contextual representations with BiLSTM-based regressor for lexical complexity prediction

Abdul Aziz, Md. Akram Hossain, Abu Nowshed Chy, Md. Zia Ullah, Masaki Aono

Abstract: Highlights•Lexical complexity prediction is a subtask of text simplification.•Fusion of transformer models provides more meaningful representations of the inputs.•The BiLSTM-Regressor improve pairwise learning between sentence and target word.•Contextual features from transformers effective for lexical complexity estimation.