The project

Development of a border security passenger screening application for an international client and border security agency.

The scope

Passenger screening and risk profiling can be unpredictable by nature and require fast response times. This information is not always accurate. Data sources are diverse e.g. immigration systems, national watchlists, traveller records/history, visas as well as external sources such as Interpol’s lost and stolen records. Corballis’ client requested us to implement a real-time and scalable passenger screening system that can be integrated with various data sources and provides response times in the milliseconds range.

The solution

We delivered a custom analytic tool which aids border security agents to analyse connections between passengers and targets in near real-time. The implemented system architecture allows border security agencies to handle hundreds of millions of passenger records each year without any impact on the system’s performance. The solution leveraged Corballis’ innovative graph visualisation library, which supported the graphical representation of people and the connections between them in the form of a graph that can be used to execute complex search queries, revealing hidden patterns and linkages. This tool allows users to search and view visualisation networks in near real-time.

Analytics services included:

  • Advanced passenger screening module based on a distributed NoSQL and graph database
  • Integration with third party data sources (watch lists, lost and stolen records, immigration systems, etc.)
  • Automatic data quality management (data cleansing, de-duplication, auto correction of mistakes, validation, etc.)
  • Advanced searching capabilities (fuzzy matching, “more like this” querying, etc.)
  • Target profile management
  • Scalable architecture

The Corballis visualisation library supported the following functions:

  • Accessing of huge amounts of data in easily digestible visuals
  • Interactive visualisations make it easy to explore and analyse large quantities of passenger data
  • Highlighting of exceptions and outliers
  • Customisable flight and passenger network visualisation
  • Intelligent layout mechanisms to make sense of data easily
  • Rich set of interactive controls
  • Zooming, panning, drag & drop support
  • Dynamic filtering of the visualisation to allow drill-down and roll up operations
  • Export the visualisations to a wide range of formats (images, JSON, PDF, etc.)

The outcome

Where profiling was once a manual and tedious process based on high level and generic rules, our solution facilitates the targeting of much more granular and accurate profiles ensuring a more streamlined and efficient screening process. The operator is no longer concerned with too much data at any one time as the system was designed to cope with hundreds of millions of passenger records in real-time.

Technology stack

  • Corballis Visualisation Library
  • Java
  • Spring Framework
  • Elasticsearch
  • Titan graph database
  • AngularJS