A novel dense retrieval framework for long document retrieval

1 Introduction Inspired by the impressive success of BERT[1]in various NLP applications,researchers have attempted to apply pretrained language models to information retrieval,and existing BERT-based retrieval models obtain improved performance on passage retrieval[2-4].Since BERT has the limitation that the maximum length of tokens is only 512,however,simply applying those models to the task of long document retrieval derives suboptimal results.The main challenge is that the direct application of BERT to long documents ignores the semantic associations among passages which are essential to eenerate accurate ranking scores.