“…As important metadata summarizing the core contents of a document, keyphrases are intended to index a document and enable readers quickly to find it and identify whether or not it is relevant to their specific needs or interests, which can also help to improve the visibility and influence of the documents (Gbur & Trumbo, 1995; Hartley & Kostoff, 2003; Turney, 2002). Meanwhile, because of their high importance and abstractness in the documents, keyphrases are also regarded as a suitable representation of topics, concepts, and knowledge, and are widely used for topic evolution study, knowledge mining, information retrieval, text summarization, and other NLP tasks (Cheng et al, 2020; Firoozeh et al, 2020; Hernandez‐Castaneda et al, 2020; Hu et al, 2019; K. Lu & Kipp, 2014; W. Lu et al, 2019, 2021; Sesagiri Raamkumar et al, 2017; Sun et al, 2020; Yang et al, 2022; Yoon et al, 2018). Owing to the importance and good performance of keyphrases regarding these tasks, automatic keyphrases prediction, including keyphrase extraction and generation, based on different algorithms, has attracted extensive attention (Çano & Bojar, 2019; Hasan & Ng, 2014).…”