Online Detection of Child Grooming Conversation

Computer Engineering

Abstract- The extensive spread proliferation of social media has unlocked up possibilities for criminals to direct the crime of online child grooming. Because of the pervasiveness of the problem scale, it can only be nominated effectively and efficiently using an automatically negotiated conversation recognition system. Previously, Gunawan, Soivito and Pranotto, [1] developed a logistic model for this purpose and the model was able to achieve 95% detection accuracy. The present study intends to solve the problem using the support vector machine and the k-nearest neighbor classifier. In addition, the study also proposes a low computational cost classification method based on the number of negotiation characteristics of the current grooming. All proposed methods are evaluated using 130 conversational texts, out of which 95 texts are prepared and 45 texts are not prepared. We recognize that there are 15 features in grooming interactions. The results suggest that SVM and K-NN are able to identify interactions at 97.6% and 96.8% of the level of accuracy. Meanwhile, the proposed simplex method has 96.8% accuracy. Empirical studies also show that between two seventeen features are unimportant for classification.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.