Applying Transformer-Based Text Summarization for Keyphrase Generation

Keyphrases are crucial for searching and systematizing scholarly documents. Most current methods keyphrase extraction aimed at the of most significant words in text. But practice, list keyphrases often includes that do not appear text explicitly. In this case, represents an abstractive summary source paper, we experiment with popular transformer-based models summarization using four benchmark datasets extraction. We compare results obtained common unsupervised supervised Our evaluation shows quite effective generating terms full-match F1-score BERTScore. However, they produce a lot absent author’s keyphrases, which makes ineffective ROUGE-1. also investigate several ordering strategies to concatenate target keyphrases. The showed choice strategy affects performance generation.