This Algorithm Monitors Twitter for Fraud in Real Time

Program to Search Twitter in Real Time- Doubtful Links in Tweets

Social media sites have now become prone to fraudsters in alluring unwary users having links to financial fraud websites. In order to avoid this, an undergraduate at Harvard University, Daniel Rothchild had created a program which tends to search Twitter in real time, automatically for doubtful links appearing in tweets.

 As per results circulated in the Journal of Technology and Science, his program discovered over 70,000 suspicious tweets within 24 hours with about 56% of the tested links considered to be deceitful. The most tweeted deceitful link to a purported weight loss program had been tweeted over 12,000 times where all of the top 10 most tweeted links were said to be fake.

A 2014 survey, a joint project by the UK government, a Get Safe Online initiative, the National Crime Agency and the telecoms regulator Ofcom and others, observed that about 51% of Britons who had been surveyed were the victims of identity theft, hacking or abuse on social media and the losses from online fraud amounted to £670m a year.

In several cases, fraudsters tend to entice the users to websites they may have set up for the purpose of collecting their personal or financial information to present them with advertisement which tend to make deceptive offers.

Social Media – Tool to Connect with Victims

Considering the massive growth in popularity of social media, fraudsters tend to use social media as new tool to connect with the victims. There are various advantages to proactively find and monitor fraudulent websites.

At first, data, collected in this manner is unbiased by whether the user of various kind of fraud intend to report or not. There are some frauds which are under reported with many targeting specific demographics and those which are reported do not necessarily represent those who seem to be heavily targeted. Secondly, fraudsters often put up, take down or move their fraudulent websites on short timescales, probably to dodge detection.

Monitoring social media data constantly could enable consumer protection groups to be conscious of fraudulent websites quicker than otherwise. Lastly having a complete record of deceitful activity on social media could show evidence to the law enforcement while pursuing fraudster trying to conceal the extent of their fraud.

However, Rothchild’s procedure tends to turn cyber-criminals’ tools against them to proactively find areas of criminal activity before they get started.

Create Browser Extension to Alert Users

In order to help user in protecting themselves, Rothchild recommend that it would be possible in creating a browser extension which would alert users when they click on a link through Twitter which tends to be automatically identified by the tool as being suspicious.

He wrote a Python script to locate for keywords in Twitter’s real time stream and the keywords were termed loosely connected with common forms of fraud. These included muscle, diet, acai, cambogia, weight, lost fast and miracle pill that were taken from a report on online fraud that was published by the US regulatory body Federal Trade Commission.

They identified many terms that could have be associated with fraudulent offer. They included diet, exercise, weight, weight-loss, prize, lottery, sweepstakes and winner.

The precise option of keywords did not matter since it could be swapped to utilise any keywords and would be suspected of being connected with fraud. The procedure searched for precise links tweeted strangely large number of times, which indicated that a bot rather than a human was behind it.