2 — Designing NLG Studio projects
As with any software project, the design stage is critical to the success of an NLG project.
For an NLG Studio project, this involves the following:
identifying the input data
deciding what (if any) data processing needs to be done, and where
deciding which data format (JSON or tabular) will be provided as input into NLG Studio
planning which scripts and variables will be used to generate the final narrative
Most important to the design stage of an NLG Studio project is that your design separates data processing and linguistic processing. Arria’s experiences show that Studio projects are much easier to develop and maintain if there is a clear separation between analytics and narrative.
Important
Make sure you separate your NLG project's data processing from its linguistic processing. In NLG Studio, this means you should have two sets of scripts: one set that contains analytics, and one set that contains the narrative. Separate the ATL code that manipulates data from the ATL code that generates the narrative's wording. This way, you can build, revise, test, and improve your Studio project more simply, efficiently, and elegantly. As an added bonus: contributors will more easily see how the project works.
Further information on designing NLG Studio projects is organized as follows:
