Annotation and Identification of Scientific Data Sharing Information from Data Availability Section

This is the authors' abstract. We don't add key points for this paper.

With the advancement of the open science movement, an increasing number of institutions and journals now require authors to explicitly state data availability in their publications, thus promoting the open sharing and accessibility of scientific data. The aim of this study is to extract scientific data sharing information from data availability statements in scientific papers. In more detail, this study annotates 8,508 data availability statements in research papers from the PLOS corpus over a period of nearly 16 months. In the end, a total of 35,010 entities and 8,524 relations covering 8 types of entities and 2 types of semantic relationships are ultimately annotated. Based on the annotated data, the model on the basis of Universal Information Extraction (UIE) is fine-tuned to automatically identify entity and relation mentions from data availability statements of the remaining scholarly articles. Experimental results show that our model is capable of extracting scientific data sharing information.