Text Categorization with Fractional Gradient Descent Support Vector Machine

Text documents on the web are an incredible resource including one example of big data, large size and so many variations that it becomes difficult for humans to choose meaningful information without the help of a computer. Text categorization job is to automatically classify text documents into standards class based on their content. The objective of this research is to implement a classifier with optimization based on the Fractional Gradient Descent in text classification. In our research, we propose using the Fractional Gradient Descent to optimize the SVM classifier so that it can increase the speed of training data. We explore a batch of different training data to compare the speed of the UCI ML text dataset training process with the SVM- SGD and SVM-FGD classifiers. This research concludes that using SVM-FGD will optimize the training time for text dataset in the activity of data classification.