Effective Trust Predictive Model for Online Advertisement Using M-Anfis Classifier

Authors

  • R. Selvavinayagam
  • A. Ashok Kumar

Abstract

Online advertising (OA) is an effectual way for a business to find new customers, expand its reach, and also to diversify the revenue streams.In real-world, lots of distrusted advertisements are presented; most researchers direct the prediction of trust advertisement, in existing work the prediction accuracy is low. To resolvesuch drawbacks, this paper proposed an effectual trust predictive model for online advertisement using M-ANFIS classifier. This proposed method consists of six steps. Initially, the data are collected from the web. Secondly, those data are preprocessed, the preprocessing phase is split into ‘3’ parts as, replacing of missing attributes, strings to number, and normalization for providing better result in the classification stage. Thirdly, some advertisement related features are extracted. After that, the best features are chosen from the extracted features using Modified Gray Wolf Optimization (MGWO). In the fifth, the advertisement is predicted as trusted and distrusted advertisement centered on the selected feature. Here, the Modified Adaptive Neuro-Fuzzy Inference System (M-ANFIS) is utilized for the purpose of classification. In the sixth, trust value is calculated centered on the click-through rates (CTRs). The proposed classifier's experimental results are contrastedto the existing entropy SVM classifier in respect of statistical measures, for instance, precision, recall, F-score, accuracy, TPR, FPR, and MCC. The proposed trust predictive model for online advertisement provides better accuracy.

Downloads

Published

2020-02-09

Issue

Section

Articles