Our client: Sage Publishing
The editors at Sage Publishing go through many papers and need to read through the papers to get an idea of whether the paper will be well received or not. If it seems like it will be, steps should be taken, such as arranging paper reviews by well-known and respected people within the appropriate field.
To help with this, predicting the citation count of a paper helps to give the editors a general idea of the paper’s success. The citation count of a paper is the number of papers that cite it as a source. The higher the number, the more important it is. The prediction of this number will not be 100% accurate, and is only meant to give a general idea.
We successfully trained models based on linear regression, random forest, and xgboost that takes in a title and abstract and predicts the citation count of the respective paper. Our key performance indicator (R^2 value) is of a satisfiable level (~0.4). We have also created a web app that connects to these models and returns a predicted citation count based on a title and a file containing the abstract.
Led data processing and experimentation and helped implement frontend.
Mainly handled implementation of backend as well as helped with data sanitization.
Mainly handled UI design and the frontend implementation.