The One-vs-Rest Method for a Multilabel Patent Classification Machine Learning Approach using a Regression Model

Redundancy in patent abstracts, that is, the repetition of an International Patent Classification tag or patent class information across documents, poses a challenge in patent classification, which is essential for efficiently managing the growing number of patent documents. This study aims to develop a model to analyze the weights and index values of summary sentences to improve the efficiency and accuracy of patent classification. This model can predict relevant patent data types. Utilizing logistic regression and the one-vs-rest method, this study focuses on the development of models to analyze the weights and index values of summary sentences, with the goal of enhancing the precision and efficiency of patent classification. The main contribution of this research is the creation of predictive models for new patent documents to streamline patent citations based on document type. This study also serves as a reference for evaluating Indonesian patent document classification models, thereby facilitating the patent evaluation process.