There are various axis on which one can classify a Data product:

  1. Readonly or mutable - does the product offer a dashboard/report (along with drill downs or rollups) or does it also let one change the system (probably like online machine learning #todo/verify)
  2. Information radiator and/or actuator - on the spectrum of whether the end user consumes it as an aid to make a decision/action, or the system itself does the action. Sort of related to above.
  3. Source aligned or consumer aligned
    1. note: this is possibly related to the idea of bounded context interaction patterns
  4. Level of abstraction - on the spectrum where consumer is partitioning/physical-storage format aware to an automated action (like fraud detection) which abstracts over multiple sub-systems. Could even goto data as a service (highest level of abstraction?)
  5. Level of determinism - on the spectrum of deterministic schema based operations to difficult to control LLMs

Tags