Intelligent Comments Filtering: A Natural Language Processing (NLP)-Based Method for Negative Review Prediction

By Fei PAN, Zefeng LIU, Haili SONG, Xiao MA, Chenchu ZHU, Mindi HUANG

Highlights

• Autonomous, real-time monitoring and filtering of malicious content

• Ensemble learning strategy leveraging multiple algorithms to enhance classification accuracy

• Meticulous text processing addressing cross-language content, imbalanced data and multi-label correlations among malicious comment types

• Feature extraction combining word frequency and importance to capture true semantic value

• Dynamic optimization enhancing filtering accuracy in response to evolving online content

Benefits

• Overcomes limitations of single-model classifiers, achieving more reliable content regulation

• Reduces manual review workload, thereby enhancing moderation efficiency

• Fosters a safer and more respectful online community