This project has documents and diagrams for design and implementation of a conversational agent named “Data Acquirer” aimed at providing code or data to data acquisition commands.

Project's goal is to make a general component design and a more or less complete grammar for the envisioned dialogs.

Here is a mind-map that outlines of the scope of the project:

Here is a mind-map that shows the components of the conversational agent:

See the corresponding Raku packages [AAr1, AAr2].

The following diagram shows one way of using DataAcquirer:

We can have a Command Line Interface (CLI) with which we can specify a chain (UNIX-like pipeline) of commands:

dsl translate -c "get a sample of 3000 JSON commodity files from AWS; 
                  parse them into long form data frame; 
                  make a data package for them; 
                  open a notebook with that data package loaded" | 
dsl data-acquire | 
dsl data-wrangle |
dsl make-notebook

The chain of four commands above:

1. Parses and interprets the natural language commands
   • And produces, say, JSON or XML records that have executable code.
2. Gets the commodities files from some default location
   • Might ask for credentials.
3. Uses some JSON parsing package in some programming language to make the long form data frame
4. Makes a data package with the long form data frame
   • Python and R data packages are regular packages.
   • WL data package can be a resource function or resource object.
5. Uploads the data package in some repository
   • Local packages installation folder or a private cloud.
6. Creates a notebook and populates with command(s) loading the data package
7. Opens the notebook in some notebook interpreter.
   • Jupyter notebooks can be opened in a Web browser or VSCode.
   • R notebooks, in IntelliJ or RStudio
   • WL/Mathematica notebooks, in Mathematica

Of course the command data-acquire might/should ask for credentials.

Similarly, all commands can take user spec, e.g. -u joedoe32.

An alternative representation of the workflow above:

dsl interpret -c "get a sample of 3000 JSON commodity files from AWS" |
dsl interpret -c "parse JSON files collection" | 
dsl interpret -c "transform into long form data frame" | 
dsl interpret -c "make a data package" |
dsl interpret -c "open a notebook with that data package loaded" 

Here is a diagram that shows the interaction between workflow steps and DAW components:

(The NLP Template Engine has its own repository, [AAr4].)

DataAcquirer is very similar to the conversational agent Sous Chef Susana in both component design and grammar design and elements.

See also the Raku package: DSL::English::FoodPreparationWorkflows, [AAr3].

[AAr1] Anton Antonov, DSL::Entity::Metadata Raku package, (2021), GitHub/antononcube.

[AAr2] Anton Antonov, DSL::English::DataAcquisitionWorkflows Raku package, (2021), GitHub/antononcube.

[AAr3] Anton Antonov, DSL::English::FoodPreparationWorkflows Raku package, (2021), GitHub/antononcube.

[AAr4] Anton Antonov, NLP Template Engine, (2021), GitHub/antononcube.