Phishing Attacks in Social Media

Many application system are been created and many research have been done to detect suspicious words messages, chat, profile, email spam, URL malicious web content also to detect phishing attacks in social media. The social approach to detect spam or malicious on Facebook, Twitter, Myspace was based on information found in social media by identifying or detecting malicious URL link, email with spam content, suspicious words from image and video text.

There’s few application that detect suspicious words embedded in images or video with help of image processing, image retrieval techniques. Previous study monitoring messages sent through social networking sites and instant messages. The designed framework that prevent, predicts, provide evidence of cyber-attacks profile when suspicious messages sent between users, but fail to detect suspicious messages in short form and coded words form sent via IM and SNM in real time. They also fail to detect suspicious words in all kind of long and short words form and coded words embedded in image contents.

Mohd Mahmood Ali, Khaja Moizuddin Mohd and Lakshmi Rajamani are researchers that found out SMD framework to detect suspicious word from messages stored in database after users made communication via social media{Ali2014}. This paper was not focus on coded words and short form chat messages. It was to prevent, predict and provide evidence of suspicious words and tracing the profile of individual or group committing crime and report to E-crime department. However the approaches used such as data mining, ontology structure divided the websites texts semantically with the assistance of Word Net database into different attributes of threats categories for example: murder, kidnap and sexual . But ontology wasn't refreshed frequently with new code words that are discovered utilizing data mining approach.

Rajamani, Lakshmi Ali, Mohammed Mahmood Rasheed, Mohammed Abdul presented a system designed with a thought of text secure framework that recognizes suspicious messages which prompts unlawful movements by criminals. A framework was not concentrated on securing messages by utilizing encryption approaches and furthermore does not focus on short form messages. This paper gives different thoughts regarding stemming calculation and priority algorithm .

Murugesan, Devi, Deepthi, Lavanya and Annie Princy. They proposed the system monitoring suspicious discussions automatically on online forum, they used text analysis to detect suspicious post in online forums. They focus on automated classification to identify more important suspicious discussion{Murugesan2016}. Thivya Shilpa. Gv proposed a framework which provide security, predicts, detect coded words and short form of suspicious words with the help of association rule mining techniques and ontology concept that provide security for the stored chat messages by using encryption technique. But this paper doesn’t detect suspicious word attached in image contents .

Salim Almi Omar Beqqali, a researcher discovered automatic system for detecting suspicious profile in social media through identifying suspicious performance and concern of users. The techniques presented is mainly based on the calculation of similarity distance to differentiate suspicious posts using text analysis. The limitations found executing time, development automated classifications and using techniques. John Resig Ankur Teredesai  in their paper explore framework for IM and various data mining issues and how they relate to Instant Messaging and current Counter-Terrorism efforts. And this paper does not tend to fully detect suspicious messages, not even detection of topics and social network analysis.

Those above articles explains approaches and limitations of capture suspicious word. Currently, clients are using different tricks of sending short form or coded words which is not so easy to recognise by crime departments or admin. Our proposed algorithms is to detect suspicious word in short form or coded words attached in images that are sent by users .