In our research, we have applied machine learning to develop a predictive model for the citation of research papers in patents. By utilizing these predictions, researchers applying for patents can determine the practical applications for their work.
We have used a Random Forest Classifier and a large set of training features to quickly predict patent citations across countries, fields of study, forms of social media, and levels of education. Data from the Altmetrics dataset provided a large corpus of material with which to train and test the model. This enhances the prediction capabilities of the model while making it generalizable to unseen data.
In this study, we have managed to achieve a rather high accuracy and recall with our model and have designed it to be easily expanded upon in the future.