Network Protection, Fraud Detection, Personalized Marketing and Real-Time Analytic Discovery
FICO (NYSE: FICO) has been awarded four new patents related to fraud detection and advanced analytics. FICO now holds 170 US and foreign patents, and has 93 pending patent applications.
Members of FICO’s fraud analytics team patented a network assurance analytic system that monitors telecommunications networks with real-time, streaming, self-calibrating analytic models. FICO uses this technology to determine abnormalities in network traffic, and the invention is also incorporated in the company’s work in cyber security.
FICO also patented a technology for “fuzzy tagging,” which involves a method for tagging a transaction for fraud risk. Rather than tagging each transaction as a binary fraud or non-fraud, the technology uses a spectrum of continuous values to tag transactions, effectively weighting them by their correspondence with fraud across the entire FICO Fraud Consortium. This technique is used in the FICO® Falcon® Platform as well as in custom modeling projects to help improve fraud detection, reduce false positives and improve the customer experience.
In terms of marketing solutions, FICO patented a “product space browser,” a graphical user interface that helps retailers discover the relationships between a customer’s purchases and their needs. The invention allows retailers to turn masses of purchase transaction data into visual maps that show product affinities and purchase patterns by customer and across customers. This facilitates the discovery of new insights, even for retailers with millions of customers and hundreds of thousands of products. The technology is one of the innovations that led to FICO® Analytic Offer Manager, FICO’s groundbreaking solution for highly personalized marketing.
FICO’s fourth new patent is for real-time determination of data stream semantics. This innovation analyzes large amounts of information in real time to categorize the information into clusters of semantic similarity without knowing the contents of the data or having a prior example of the data structure. This invention can be useful in many applications where new datasets are received that must be classified instantly.
Source: FICO Press Release
