Detection Packs 0.2 Release

Greg_Jones

We are excited to announce the release of version 0.2 of Detection Packs. This is our first major upgrade to Detection Packs since 0.1, which contained the Generic FinCrime Detection Pack MVP. For full details of the release, including compatible Quantexa Platform versions and minor enhancements, please see the Quantexa Documentation site.

Deploy Detection Packs Scoring quickly and easily using our new low-code interface

As part of the simplification journey of the Quantexa Platform, this release of Detection Packs introduces the low-code deployment of Detection Packs Scoring. The previous 0.1 release provided a code-based interface for deploying and configuring the out-of-the-box Scores, Scoring pipelines, data mappings and data providers. All these base components are now set up using configuration files, meaning a significant reduction in the required Scala and Assess knowledge to deploy, and in turn a reduction in both the cost of implementation and the time to value for our clients.

A new Detection Packs plugin provided with this release takes the configuration files and generates the required Assess-based Detection Packs Scoring code, in the same manner as our Data Fusion product does for ETL code. The Detection Packs configuration files and plugin are designed to be familiar to existing users of Data Fusion as well, utilizing the same format and usage patterns, respectively. The configuration-based data mappings in Detection Packs 0.2 validate directly against your Data Fusion model files, and further closer integration is planned in future.

Example showing how a Score in version 0.1 (EntityDirectLinkToWatchlist.scala) is now represented by a simple config file in version 0.2 (EntityDirectLinkToWatchlist.qscore)

Although you can now set up and deploy all the built-in Detection Pack Scores without writing a single line of code, we have retained the flexibility for users to add custom Scores for their specific deployment. This has been enabled by the introduction of a number of entry points into the Detection Packs generated Scoring pipelines, allowing users to achieve the benefits of low-code, configuration-driven Scoring whilst still being able to plug-in their own custom code-based Scores.

This is all the more powerful for our ever-expanding collection of Reference Scores. Reference Scores are pre-written Scores created in conjunction with our users to provide additional Scores over and above the core Detection Pack for FinCrime. They also cover additional use cases outside of FinCrime, and the constantly growing catalog currently contains over 60 Scores. The Scores themselves are designed, built, and reviewed so that they can be easily plugged into the Detection Packs Scoring pipelines as-is, but also so they are ready to be adopted in future as official Detection Packs Scores released in a configuration-driven format.

Efficiently utilize Detection Packs for Capital Markets and Retail Banking with added native support

This release provides native support for two additional use cases – Capital Markets and Retail Banking. These use cases both require a different scoring pipeline to our previously natively supported Trade Finance use case, and so have been enabled by the introduction of use case specific Scoring pipeline generation in our new Detection Packs plugin.

A user simply has to define in a configuration file which use case their deployment needs, and the plugin will generate the relevant scoring pipeline. Other use cases can still benefit from Detection Packs by using one of our natively supported pipelines as a baseline and then making customizations to the generated code as required.

Example qsolution configuration file showing how a user can now easily define their use case as Retail Banking

Coming soon to Detection Packs

We are currently targeting early 2024 for the 0.3 release of Detection Packs, with the expectation of releasing our version 1.0 within the next financial year. Here are some of the planned features our users can look forward to:

• Adoption of many more Reference Scores into officially supported, configuration-driven Detection Packs Scores
• Further additional use case support
• Dynamic pipeline generation
• Score versioning and seamless upgrade support
• Improved display of Scores and Scorecards in the UI
• Low-code, configuration-driven Task Loading