Detection Analysis Of URL Attack In SDR Systems For Network Data Security Using Machine Learning Methods

Abstract

During day-to-day life, most of the people using internet for net banking, shopping, etc which provides risk to the user from internet.  Spoofing (Phishing) means the appearance of treachery wherein the aggressor endeavor to discover sympathetic information such as users access code or account details through conveyance it to either electronic mail or some more communication lines as a trustworthy individual or someone. In general, the affected party got statement which emerges to be mailed by recognized person or association. This statement consists of either malevolent software focused the user’s desktop or having associations of personal endure to malicious websites with the intention of scam them into revealing private information namely account ID, credit card, password details. Now, lack of security in the data network caused if aspiration to analyze with big data, Artificial Intelligence and Internet of Things. In such type of situation, URL phishing attacks in the website have to be found through statistical analysis in the proposed study.  Such type of phishing URL has to be classifying either as phishing URL or non-phishing URL via statistical analysis in machine learning techniques.  Such type of classification methods can be done through data pre-processing, feature extraction, training, testing and validation phase of machine learning techniques. This proposed manuscript deals with data mining classification methods like decision tree classifier, K- Nearest Neighbor classifier and ensemble gradient boosting classifier also to classify the phishing URL as malicious or normal for various features of URL and websites. Thereby, the enhancement of overall model performance can be evaluated by finding metrics as accuracy, precision and recall.