Fake Website Detection from the Perspective of Network Security: Preventing Counterfeit Brands in New Media Marketing

Abstract

With the popularization of new media marketing, fake websites and counterfeit brands are becoming more rampant, posing huge risks to consumers and businesses. Traditional website monitoring has obvious delays when processing large numbers of web pages, and the data source is also relatively single and relies on specific platforms, resulting in insufficient sample data and feature deviations. Therefore, this paper uses an improved Transformer model to construct an intelligent fake website detection model. By collecting sample data from fake and legitimate websites on different platforms, data labeling is performed to provide high-quality data for model training. Deep learning technology is used to extract website features, and indicators that best represent the features of fake websites are selected to enhance the accuracy of the model. An improved Transformer model is constructed, and model parameters are optimized through a large number of sample training, improving its identification speed of fake websites. Model transfer learning is designed, and model fine-tuning and domain adaptation strategy technology are used to improve the adaptability of the model on data from different platforms so as to ensure that the identification of fake websites is not limited to the data characteristics of a single platform. Experimental results show that the optimized Transformer model maintains an accuracy rate of over 97% when identifying fake websites on the commonly used browsers Google Chrome, Mozilla Firefox, Microsoft Edge, and Safari platform, and the F1-score on these four platforms is as high as 0.96, 0.97, 0.98, and 0.98, respectively, demonstrating its good classification performance. Although the training time of the model is slightly longer, it is still within an acceptable range and does not negatively affect the overall performance of the model. In terms of detection time, the classification process only takes 27 seconds to 33 seconds, indicating that it has a fast processing speed. The improved Transformer model studied has potential application prospects in the intelligent detection and identification of fake websites from the perspective of network security.