Classification of Patents according to Industry 4.0 Pillars using Machine Learning Algorithms

Abstract

Industry 4.0 is on the horizon. Therefore, it is crucial to analyze the patterns and trends of intellectual property (IP) information to determine the readiness of stakeholders to adapt to the changing industrial evolution. Patent bibliography documents consist of structured and unstructured data, so text mining or machine learning must be employed for the data analysis. This paper established a patent trend by analyzing the patent data of Intellectual Property Corporations of Malaysia (MyIPO) to identify the institution’s readiness to face the fourth industrial revolution. To achieve this aim, a patent classification method was used to classify MyIPO patent data based on the pillars of Industry 4.0. Furthermore, the patents data were drawn from MyIPO Online Search and Filing System was used as the datasets in this study. However, the dataset consists of the title of the patent and the publication year only. Since short text data in the title has fewer semantic information and high sparseness, this issue was a challenge for this study. In this paper, five common classifiers were used for text classification. Support Vector Machine (SVM) was proven to be the machine learning classifier with the highest accuracy in classifying the training and testing datasets. The findings of this paper present the patent trend for each pillar of Industry 4.0 including the patents related to Industry 4.0 where Autonomous Robot is the pillar with the highest innovation.