Got a Complaint?- Keep Calm and Tweet It!

摘要

Research shows that many public service agencies use Twitter to share information and reach out to the public. Recently, Twitter is also being used as a platform to collect complaints from citizens and resolve them in an efficient time and manner. However, due to the dynamic nature of the website and presence of free-form-text, manual identification of complaint posts is overwhelmingly impractical. We formulate the problem of complaint identification as an ensemble classification problem. We perform several text enrichment processes such as hashtag expansion, spell correction and slang conversion on raw tweets for identifying linguistic features. We implement a one-class SVM classification and evaluate the performance of various kernel functions for identifying complaint tweets. Our result shows that linear kernel SVM outperforms polynomial and RBF kernel functions and the proposed approach classifies the complaint tweets with an overall precision of 76 %. We boost the accuracy of our approach by performing an ensemble on all three kernels. Result shows that one-class parallel ensemble SVM classifier outperforms cascaded ensemble learning with a margin of approximately 20%. By comparing the performance of each kernel against ensemble classifier, we provide an efficient method to classify complaint reports.