NetReveal analytics platform

Financial Crime 4 Март 2015, 17:07

Risk based targeting of financial crime underpinned by analytics.

Our NetReveal® platform embodies more than a decade of research and development investment and provides a suite of specialised data ingestion and transformation tools, analytics, risk scoring, simulation and model management capabilities - all of which are optimised for fraud, risk and compliance solutions.

Our experience in delivering real-world solutions in complex and changing environments is the backbone of our NetReveal analytics philosophy. Rapid time-to-value is achieved through deployment of out-of-the-box rule libraries; as more data becomes available, sophisticated modelling techniques can be rolled out to significantly reduce false positive rates. All techniques are white-box, providing the transparency essential in an increasingly audited and regulated climate.

Detection techniques are tailored for each specific domain. For example, while transaction fraud models can leverage known fraud cases, detection of unauthorised trading requires unsupervised learning and corporate fraud solutions require text analysis; compliance solutions use rule libraries derived from regulatory principles coupled with statistical profiling and matching algorithms; insurance and tax fraud solutions use network models to identify high-value collusive patterns.

Our NetReveal analytics platform supports all of the following and techniques can be chained together to implement hybrid approaches.

– Behavioural profiling - pre-aggregating activity across multiple data-streams and multiple time scales to build up a picture of normal and abnormal behaviour. Improves detection accuracy compared to working with raw data;
– Data matching and text analysis - algorithms with options for numerous languages and character sets, address parsing tools and specialised methods for matching names to watch lists;
– Rule-based analysis - pre-packaged, standardised and tested libraries and scorecards for a wide range of fraud, compliance and risk typologies derived from our extensive experience of working with clients around the globe;
– Unsupervised learning - ideal for problems where there is no training data available, but where there is a natural structure in the data which can be discovered through an automated data mining process, and exploited to detect subtle anomalies while maintaining low false positive rates;
– Supervised models - deliver improved detection results taking advantage of a marked data population for rapid model development with the ability to re-train risk models as the underlying data drifts over time;
– Network analytics - automated entity linking across the entire data population into natural networks, and with a scoring model to highlight fraud rings, collusive networks or complex money laundering patterns;

– In the real world, data is not perfect. Robust, universal identifiers do not exist - especially where IT systems are fragmented. Our NetReveal analytics capability overcomes data quality issues, performing accurate entity resolution even where there are deliberate attempts to evade detection through falsification of information. NetReveal provides a single view to 'join the dots' across source systems and service delivery channels;

– NetReveal is open and extensible enabling clients to take ownership of all model management, rules configuration and validation tasks without requiring vendor engagement;

– The analytics platform can be deployed in batch and real-time, enabling clients to move from after the fact risk detection to fully inline risk prevention.