Alert Manager App for e-Commerce Customers

Abstract

In the electronic world we live in, where more and more commercial transactions happen over the internet and where the magnitude of losses to fraud increases every day, the need for sophisticated systems for fraud detection and prevention arise. One of the pitfalls of existing system is the inability to make sense of the extensive amount of business data available. Signi cant insights could be obtained by identifying relationships among that data. Entity Link analysis emerge as a powerful ally of fraud detection and prevention systems. By exploring relationships within a network of related entities, abnormal patterns of behaviour can be uncovered. In this report we describe the development of an exploratory tool for Entity Linking. This web application provides di erent data visualizations to analyse commercial transactions and identify relationships between them, in order to detect fraudulent entities. Data visualizations available are: table, matrix diagram, geographical referencing, force directed graph, circular diagram and chord diagram. Through usability evaluation, force directed graph was elected as the most promising visualization for fraud analysis. But both the matrix diagram and the circular diagram obtained good reviews, opening the discussion for a combined use of data visualizations to be integrated with the fraud detection and prevention solution.