An adaptive neuro-fuzzy inference system for detecting redirection spam

Redirection spam detection is important to maintain the integrity of information on World Wide Web. It is of prime interest to search engines for quality information retrieval. Malicious redirections break the trust of users and create unusual traffic leading to wastage of expensive bandwidth and other resources. But detecting redirections is complicated owing to the genuine use of redirections for load balancing and URL shortening. Present work addresses this problem of malicious redirections and proposes an adaptive neuro-fuzzy model (ANFIS) for its detection. The model takes five input features and uses hybrid algorithm for learning. We also employ subtractive clustering technique to reduce the training time so as to facilitate the quick decision making. The proposed model is tested and validated using datasets. The experimental results indicate that the model detects redirection spam with high accuracy. Therefore, the proposed model can be used as an effective approach for redirection spam detection.