Research On Analysis And Identification Of Fake Behaviors In Academic Journal Websites Based On Measurement

The rapid development and popularization of the Internet technology have brought great convenience to people's life.The transactions in the network are becoming more and more frequent.At the same time,it provides a platform for Internet crimes,such as the phishers use phishing crime phenomenon is increasing rapidly.It not only cbrings serious economic loss and serious challenges of social trust,but also disturbs the order of network.In recent years,while searching the journal submission websites in the Internet,there are many results after entering the journal titles.But only one of them is real academic journal submission site,and the rest are contribute fake websites.However,the researchers have not yet devised a complete system of defense in academic journals.According to this phenomenon of the network,in the field of academic journal website testing,a fake academic journal websites testing technology is proposed in this papers.This method can effectively protect user privacy and maintain the order of the academic community.By studying and analyzing the anti-phishing detection technology at home and abroad,this paper proposes a kind of fake websites detection technology by combining with the characteristics of the academic journal submission websites.In this method,we first crawl the URL of all real and fake websites from the search engine by using the names of the journals as the keywords.And then we retrieve and obtain the characteristics of the URL,web page contents characteristics and the domain information features by analytical tools and the whois.Next,we count and analyze the characteristics of extraction.By the differences of the features between the real and the fake web pages,we can select the appropriate eigenvalues and calculate the weights of these eigenvalues.Finally we use the SVM(Support Vector Machine)classification learning algorithm to train classifier and then use the classifier to classify academic journals.For the different frequencies of each eigenvalue in the real and fake websites,this paper proposes a more effective method replacing the traditional boolean type numbers.By calculating the frequencies of each eigenvalue in the real and fake web pages,we can give them different weights,so that it can be effective to reflect each eigenvalue in the importance of the testing.The experiment shows that the method of labeling each eigenvalue with its weight is more accurate than using boolean type number in testing.