Is the Impact of Management Research Predictable Through the Title? - A BERT Model to Find a Response

In academia, the impact a research paper can generate is a matter of concern to most researchers. Therefore, in this study a model is proposed to evaluate whether the impact is predictive by considering the title of the article. To measure this impact, the number of times an article is cited is taken into account. In addition, the aim is to create a tool that, when a new article title is introduced, will go through the designed model and output the impact it will have in five years’ time. This paper focuses specifically on the management research field, so a dataset has been created with data downloaded belonging to this specific domain. This dataset has been labeled, preprocessed, tokenized, padded, masked and split into training and validation sets. The data were then trained and evaluated across a BERT model. The F1-score performance metric achieved is 0.56. Finally, some possible improvements are proposed.