Now that you have a built classification model, we need to implement an operational topology that leverages this model in order to perform classification as part of a larger operational data pipeline. I would like to draw a distinction between an operational data pipeline and an operational process. I will talk about operational process as an architecture concern that involves potentially many system layers. These may include ERP, CRM, core processing engine, and so on. An operational process is, therefore, positioned at the solution architecture level, and, as stated in the previous recipe, it typically won't include an analytics platform, such as R. A data pipeline is applicable at Trident's level of abstraction. Trident, in effect, allows you to define a streaming data pipeline. It abstracts away that state and planning in order to achieve this pipeline at scale and in parallel.

This recipe will therefore present operational...