Identification of ‘Early Career Stage’ Researchers from large scale Bibliographic dataset

In the realm of scientific research, collaboration plays a pivotal role. Identifying and supporting Early Career-stage Researchers (ECRs) is crucial for fostering their career development and ensuring the growth of scientific knowledge in their respective fields. In this paper, we present a machine learning-based approach to identify ECRs from large-scale bibliographic datasets. Our proposed framework leverages various features extracted from the dataset, including publication records, citation patterns, academic affiliations, co-/authorship networks, etc. To train and evaluate our model, we utilize a comprehensive and diverse AMiner Author dataset. Further, we have divided the 5000 sample data points into a 7:3 ratio for training and testing purposes. However, with the application of machine learning techniques, we achieve high accuracy in identifying ECRs, effectively distinguishing them from established researchers (‘h-index’ distributions are used to identify the established researchers).