Sidestep redirection spam with fuzzy logic

Web browsers might soon use fuzzy logic to spot redirection spam and save users from being scammed, phished or opening malicious sites unwittingly, according to researchers in India writing in the International Journal of Electronic Security and Digital Forensics.

Redirection spam occurs when a user opens a link in an email that leads to an unexpected and often malicious page, or when they open a page that has been hacked or injected with malware, which then redirects to a malicious page. Often the redirection occurs instantaneously and transparently without the user being aware until it is too late and login details or credit card number have been divulged to the criminal third party. Frequently, there will be a malware payload that infects the user’s computer at the same time.

According to Kanchan Hans of Amity University, in Noida, India, and colleagues, legitimate web page redirections are a ubiquitous part of the web used for server load balancing, link logging and URL rewriting and shortening. Detection of illicit redirections is a difficult task as blocking them would block legitimate redirections too. Nevertheless, while redirection spam was originally little more than a “switch and bait” black hat search engine optimization (SEO) technique, today it interferes with the performance of search engines, leads to wastage of network bandwidth and disrupts user trust, as well as leading to fraudulent activity, identity theft and the spread of malware.

Most modern web browsers have security tools in place that will alert the user to the presence of malware on a site they attempt to visit. Unfortunately, this relies on the developers of the browser having access to a continuously updated database of flagged sites. If a site has not yet been flagged as malicious, the unwary user may stumble on to a page and be the victim of a wide range of scams and problems. Hans and colleagues have developed a system that could be used in conjunction with such conventional alert systems and provide an extra layer of security against redirection spam.

The team’s detection system analyses the characteristics of a given web address based on known spammy links and applies fuzzy logic to add a layer of probability to whether or not the suspicious link is likely to be a problem. Various different criteria are applied in terms of whether the link to be followed might be spam including the number of redirection hops that would take place after the user clicks or enters an address, the presence of a refresh delay, whether or not there are JavaScript redirects on the page, whether there is a meta tag redirection in place. All such characteristics are exploited by spammers to mask the true destination of a link from anti-malware and other security tools used by browsers and the search engines and so avoid the true destination site being detected and flagged as malware.

The application of fuzzy logic allows a probability to be calculated with looser rules based on the different criteria, so that a confidence level can be assigned to a given link as to whether it is safe or spam. In an actual browser implementation this might give users a red, amber or green signal to let them know whether they should proceed to visit a site. In practice, only red and amber sites would generate an alert, sites given the green light could be set to open and so reduce the need for users to make a decision when a site is almost certainly safe to visit, but give them a chance to think twice before visit a putatively hazardous page. Tests on the system show a high level of accuracy in flagging safe and spam sites from a known database without significant false positives or negatives, the team reports.

Hans, K., Ahuja, L. and Muttoo, S.K. (2016) ‘A fuzzy logic approach for detecting redirection spam’, Int. J. Electronic Security and Digital Forensics, Vol. 8, No. 3, pp.191–204.

Research Picks Extra – June 2016

Stitching together a HEMS

Researchers in Canada and China are stitching together an intelligent home energy management system (HEMS) that reduces the number of sensors needed to monitor the domestic environment. Their MinNet system uses convenient plug-in power sensors to construct a wireless sensor network for data collection and establish an analytical framework based on the power model of appliances, the team reports. It can then be used to monitor the status of individual home appliances using the minimum number of power sensors and also estimate occupancy of a given room or the house as a whole using with inference intelligence and so optimise electrical and energy efficiency in terms of lighting, heating, air conditioning.

Tang, J.Y., Tang, G.M. and Wu, K. (2016) ‘MinNet: toward more intelligent smart home energy management systems with fewer sensors’, Int. J. Sensor Networks, Vol. 20, No. 4, pp.252–263.

When your car talks back

Engineers at the Hyundai Motor Company in Gyeongg-Do, South Korea, and colleagues are investigating how drivers might respond to voice alerts from their vehicle’s automated braking systems. Their simulations suggest that drivers are much happier when their car tells them of its automated actions rather than simply carrying out those actions for the benefit of driver and passenger safety. “Car makers should also focus on the human–machine interaction, i.e., on how the car announces its ‘intentions’ to act,” the team reports. The work provides insights into how the transition between fully driver-controlled cars and self-driving cars might take place. “Voice warnings can significantly alleviate anxiety, increase alertness and give back the sense of control to drivers by appropriately providing information about the car’s operating status,” the team says. They suggest that drivers feel safer with the voice alert system activated than when it is disabled.

Koo, J., Shin, D., Steinert, M. and Leifer, L. (2016) ‘Understanding driver responses to voice alerts of autonomous car operations’, Int. J. Vehicle Design, Vol. 70, No. 4, pp.377–392

IT for ADHD at Uni

Attention deficit hyperactivity disorder (ADHD) is thought to affect more than 1 in 10 of students at university. Now, researchers from Australia and Saudi Arabia have assessed the possibility of using information communication technology (ICT) solution to help with the current treatment and management of the condition. The focus of their work and case study involves the use of ICT systems as an effective alternative to standard pharmaceutical interventions.

Binhadyan, B. and Wickramasinghe, N. (2016) ‘An investigation of the benefit of using IT in the context of university students with ADHD’, Int. J. Networking and Virtual Organisations, Vol. 16, No. 2, pp.204–220.

Gene selection

Researchers in India have developed a new approach to analysing cancer genetic data that allows them to extract the most important genes from complex arrays and to put them into rank order. Accurate sample class prediction of gene expression profiles is of great importance in the field of medical data mining to improve the diagnosis and treatment of diseases, the team points out. Their classification approach involves “dimensionality” reduction which allows meaningful information to be extracted more effectively and quickly. In order to In order to develop their rule-based algorithm, they overcame various problems using rough set theory, decision tree algorithm, database technology, and other mathematical and statistical approaches.

Das, S. and Das, A.K. (2016) ‘Gene selection and decision tree based classification for cancerous sample detection’, Int. J. Biomedical Engineering and Technology, Vol. 21, No. 1, pp.1–14.

No ticket to ride

Urban ridesharing services could be much more effective, efficient and have greater uptake if they were more dynamic and took into account all the major factors associated with running such as a service, according to research published in Progress in Industrial Ecology – An International Journal.

Giovanni Zenezini of the Department of Management and Production Engineering at the Politecnico di Torino, and colleagues there and in the research labs of Telecom Italia explain that there are four major considerations involved in ride sharing services. The first is that a third-party provides a matching platform managing the interactions between the two users, driver and passenger. Secondly, the ride-sharing platforms make one-time ride matches that are generally not repeated and take place at random points and at random times. Thirdly, a ridesharing system must be flexible in order to match users at very short notice. Finally, the purpose of the trip being taken is irrelevant to the match.

The team points out that services such as Uber and Lyft match users at short notice and make one-time ride matches, but in reality, they directly match drivers with passengers, without taking into consideration the trip accommodation of either party. In this sense, they satisfy the first three consideration but do not take into account the fourth and so cannot be considered to be dynamic ridesharing system by definition.

Ride-sharing has the potential to reduce urban pollution and emissions and vehicular congestion, nudging cities and towns towards a more sustainable transport solution. These systems exploit the near ubiquity of smart phones and location services, such as GPS in many cities around the world to help drivers and passengers arrange shared rides and broker information exchange and remuneration as appropriate. Ride-sharing schemes face a “bootstrap” problem in that they need a good supply of drivers to encourage passengers to use the service but in order to recruit people to be drivers there needs to be sufficient demand, which means encouraging passengers to use such a service.

“Ridesharing services need to reach a critical number of users, especially in the initial phase, in order to provide a desirable level of matching between drivers and passengers,” the researchers explain. Only then can they pull themselves up by their bootstraps. The team’s case study reveals that essentially an initial boost from advertising is critical to whether or not a new ridesharing service can become viable. Subsequently, penetration rate depends on rapid growth of the driver and passenger communities. Pushing the benefits to future drivers is also paramount as is providing a backup service for passengers as and when the primary service fails to find a suitable journey match.

Zenezini points out that, “Consolidating demand can be achieved when selected group of individuals have the same destination, as in the case of employees of one company or maybe for a special event (e.g. a concert). This means that this selected group of users also share the purpose of the journey, such as commuting or leisure.”

De Marco, A., Giannantonio, R. and Zenezini, G. (2015) ‘The diffusion mechanisms of dynamic ridesharing services’, Progress in Industrial Ecology – An International Journal, Vol. 9, No. 4, pp.408–432.