Computer Science And Technology - Research Publications
Permanent URI for this collectionhttps://kr.cup.edu.in/handle/32116/82
Browse
4 results
Search Results
Item Detection of phishing websites using C4.5 data mining algorithm(Institute of Electrical and Electronics Engineers Inc., 2018) Priya, A.; Meenakshi, E.Phishing sites are fake sites that are made by deceptive persons which are copy of genuine sites. These websites look like an official website of any company such as bank, institute, etc. The main aim of phishing is that to steal sensitive information of user such as password, username, pin number, etc. Victims of phishing attacks may uncover their money related delicate data to the attackers who may utilize this data for budgetary and criminal exercises. Different technical and non-technical approaches have been proposed to identify phishing sites. Non-Technical approach has no solution against the fast disappearance feature of phishing websites. Data mining technique, one of the classifications of technical approach, has shown promising results in detection of phishing websites. As compared to non-technical approaches, data mining techniques can generate classification models which can make prediction on phishing websites in real-time. In this paper analysis of C4.5 (J48) data mining algorithm has been done using WEKA tool. C4.5 is a benchmark data mining technique which can accurately identify phishing websites. A training dataset of 750 URLs has been made to train the algorithm J48, which is an implementation of C4.5 algorithm in WEKA. Testing dataset of 300 URLs is used to make prediction using the classifier generated after the training of J48. True positive rate, True negative rate, False positive rate, False negative rate, Success rate, Error rate and Accuracy are calculated after testing process. Result shows C4.5 has an accuracy of 82.6%. ? 2017 IEEE.Item Detection of malicious URLs in big data using RIPPER algorithm(Institute of Electrical and Electronics Engineers Inc., 2018) Thakur, S.; Meenakshi, E.; Priya, A.'Big Data' is the term that describes a large amount of datasets. Datasets like web logs, call records, medical records, military surveillance, photography archives, etc. are often so large and complex, and as the data is stored in Big Data in the form of both structured and unstructured therefore, big data cannot be processed using database queries like SQL queries. In big data, malicious URLs have become a station for internet criminal activities such as drive-by-download, information warfare, spamming and phishing. Malicious URLs detection techniques can be classified into Non-Machine Learning (e.g. blacklisting) and Machine learning approach (e.g. data mining techniques). Data mining helps in the analysis of large and complex datasets in order to detect common patterns or learn new things. Big data is the collection of large and complex datasets and the processing of these datasets can be done either by using tool like Hadoop or data mining algorithms. Data mining techniques can generate classification models which is used to manage data, modelling of data that helps to make prediction about whether it is malicious or legitimate. In this paper analysis of RIPPER i.e. JRip data mining algorithm has been done using WEKA tool. A training dataset of 6000 URLs has been made to train the JRip algorithm which is an implementation of RIPPER algorithm in WEKA. Training dataset will generate a model which is used to predict the testing dataset of 1050 URLs. Accuracy are calculated after testing process. Result shows JRip has an accuracy of 82%. ? 2017 IEEE.Item Combinational feature selection approach for network intrusion detection system(Institute of Electrical and Electronics Engineers Inc., 2015) Garg, T.; Kumar, Y.In the era of digital world, the computer networks are receiving multidimensional advancements. Due to these advancements more and more services are available for malicious exploitation. New vulnerabilities are found from common programs and even on vulnerability in a single computer might compromise the network of an entire company. There are two parallel ways to address this threat. The first way is to ensure that a computer doesn't have any known security vulnerabilities, before allowing it to the network it has access rights. The other way, is to use an Intrusion Detection System. IDSs concentrate on detecting malicious network traffic, such as packets that would exploit known security vulnerability. Generally the intrusions are detected by analyzing 41 attributes from the intrusion detection dataset. In this work we tried to reduce the number of attributes by using various ranking based feature selection techniques and evaluation has been done using ten classification algorithms that I have evaluated most important. So that the intrusions can be detected accurately in short period of time. Then the combinations of the six reduced feature sets have been made using Boolean AND operator. Then their performance has been analyzed using 10 classification algorithms. Finally the top ten combinations of feature selection have been evaluated among 1585 unique combinations. Combination of Symmetric and Gain Ratio while considering top 15 attributes has highest performance. ? 2014 IEEE.Item Comparison of classification techniques for intrusion detection dataset using WEKA(Institute of Electrical and Electronics Engineers Inc., 2014) Garg, T.; Khurana, S.S.As the network based applications are growing rapidly, the network security mechanisms require more attention to improve speed and precision. The ever evolving new intrusion types pose a serious threat to network security. Although numerous network security tools have been developed, yet the fast growth of intrusive activities is still a serious issue. Intrusion detection systems (IDSs) are used to detect intrusive activities on the network. Machine learning and classification algorithms help to design 'Intrusion Detection Models' which can classify the network traffic into intrusive or normal traffic. In this paper we present the comparative performance of NSL-KDD based data set compatible classification algorithms. These classifiers have been evaluated in WEKA (Waikato Environment for Knowledge Analysis) environment using 41 attributes. Around 94,000 instances from complete KDD dataset have been included in the training data set and over 48,000 instances have been included in the testing data set. Garrett's Ranking Technique has been applied to rank different classifiers according to their performance. Rotation Forest classification approach outperformed the rest. ? 2014 IEEE.