Improving Text Classification Performance through Data Labeling Adjustment

The data labeling process in natural language processing (NLP) technology is an essential process for identifying raw text data and adding information labels to provide context for machines to learn. However, labeling set of data is one of the most time-consuming parts of any machine learning application, and requires to verify the accuracy of labels and adjusts them correctly as necessary. In this paper, we propose a method to adjust data labels to better fit the meaning of sentences to improve model performance. In addition, we show that our approach is feasible to improve model performance through various experiments