A Simple Classifier for Detecting Online Child Grooming Conversation

Abstract

The massive proliferation of social media has opened possibilities for the perpetrator conducting the crime of online child grooming. Because the pervasiveness of the problem scale, it may only be tamed effectively and efficiently by using an automatic grooming conversation detection system. The current study intends to address the issue by using Support Vector Machine and k-nearest neighbors’ classifiers. Besides, the study also proposes a low-computational cost classification method, which classifies a conversation using the number of the existing grooming conversation characteristics. All proposed methods are evaluated using 150 textual conversations of which 105 are grooming, and 45 are non-grooming. We identify that grooming conversations possess 17 features of grooming characteristics. The results suggest that the SVM and k-NN can identify grooming conversations at 98.6% and 97.8% of the level of accuracy. Meanwhile, the proposed simple method has 96.8% accuracy. The empirical study also suggests that two among the seventeen characteristics are insignificant for the classification.