Fraud management is a real issue on the web. However, not just on the web but quite pervasive in our society which takes on many forms. However, many systems are not smart enough these days to detect fraud. They are also limiting in their capacities to discriminate. Unfortunately, people often seem to take the misguided approach of trying to detect fraud in all the wrong ways especially when they attempt to discriminate on basis of racial profiling. This often adds to overfitting or underfitting of the data. Sanctions databases are also centrally managed which are at times not sufficient enough to be kept up-to-date in real-time nor can they be fully trusted for accuracy. Even credit reference agencies are not always sufficiently up-to-date with their records, in real-time, and at times can make mistakes which can effect individuals for years. How do you protect people effectively that make transactions online and take the necessary steps to protect their identities. Banks are very incompetent in protecting customers from identity fraud. Intelligent methods these days need to do more to tackle such attempts. If someone loses their life savings through fraud it should be realistically possible to get the money back, all of it, and the systems corrected for future preventative measures. However, to a certain degree this is possible today in certain regions. Although, this is a security issue in many respects, it is also a compliance issue. Fraud really needs to be viewed from an objective sense without effecting the data with subjective human stereotypes. That is the only way fraud can be more effectively detected and prevented. Also, by just detecting fraud on basis of a boolean yes or no is insufficient. Even the idea of presumptuous scoring is insufficient as it is all very subjective. Intelligent methods against fraud need to take a holistic approach. Such systems need to be embedded with what ethics really means. They need to be embedded with notion of regulation and compliance based on real-time updates. The systems could monitor transaction flows in a fully encrypted fashion. The decentralization of fraud management is good because even such services can come under disrepute. Often, fraud management and providers of such management required to maintain audits and compliance in order to provide high levels of trust. But, how often do we find carelessness with data management. While customers are victimized online, fraudsters continue to find better evasion methods and at times are coy enough to continue the same practices undetected. This does indicate that fraud management in many services especially for the web is not working and has not reached a significant crossroads of improvement. Although, it does seem valid that machine learning is the answer to most such fraud management approaches. And, detecting patterns and learning from such patterns certainly helps in targeting and retargeting. However, a richer sense of semantics is necessary as to the meaning of fraud as well as the meaning of trust. It is also necessary to provide a more secure form of distributed identity management where people could check their rolling identities and be more mobile with them, trust in the knowledge that it is intact from preying eyes. Unfortunately, even the notion of securing systems these days is rather a shady area, as there are always backdoors necessary in order to provide checks and balances especially for legal and regulatory which can inevitably provide for leaks. Intelligent systems with semantic web is the answer to most things of today. Using both statistical techniques as well as artificial intelligence that can be framed in a rich knowledge representation is a workable approach. Also, once the knowledge representation is there a more likely knowledge discovery graph can be anticipated. Most such solutions can be tracked using a graph database to understand the familial and detect richer patterns from even assisted metadata. Even approaches of applying deep learning by way of neural networks can be a positive step.
