Mobile phones have become an essential device for accessing the web. This is due to the advantages of portability, lower cost and ease. However, the adoption of mobile phones for online activities is now being challenged by myriads of cybercrimes. One of such crimes is phishing attack. In this work, a lightweight anti-phishing technique is proposed to combat phishing attacks on mobile devices. This is necessary because these mobile platforms have increased the attack surface for phishers while diminishing the effectiveness of existing countermeasures. The proposed approach uses a number of URL behavior to determine the status of a website based on frequency analysis of extracted phishing features from PhishTank. To increase the detection power of unknown pattern, a machine learning algorithm called Support Vector Machine is adopted. The results indicated that the approach is very efficient against phishing sites with negligible false negatives.

The reality of human existence and their interactions with various things that surround them reveal that the world is imprecise, incomplete, vague, and even sometimes indeterminate. Neutrosophic logic is the only theory that attempts to unify all previous logics in the same global theoretical framework. Extracting data from a similar environment is becoming a problem as the volume of data keeps growing day-in and day-out. This chapter proposes a new neutrosophic string similarity measure based on the longest common subsequence (LCS) to address uncertainty in string information search. This new method has been compared with four other existing classical string similarity measure using wordlist as data set. The analyses show the performance of proposed neutrosophic similarity measure to be better than the existing in information retrieval task as the evaluation is based on precision, recall, highest false match, lowest true match, and separation.

No abstract