Raku for Prediction

Anton Antonov — 45 minutes

In this presentation we discuss the architectural design and implementation of a software system for the specification, navigation, and utilization of computational workflows in the fields of data science, machine learning, and scientific computing.

More specifically in the presentation we are going to:

1. Introduce a simple, operational, and extendable system of natural Domain Specific Languages (DSLs) for the specification of computational workflows

2. Outline a general strategy and a software architecture of a system that translates sequences of sentences of those natural DSLs into executable code for different programming languages (and packages in them)

3. Discuss and demonstrate a Raku implementation of such software system and how to utilize it through Raku

The system we describe in the presentation is multi-language from both natural language and programming language perspectives. We give a large number of illustrating examples of its functionalities, scope, and principles. We also demonstrate a "server solution" that uses the Cro library. At the end we discuss alternative approaches, system's current state, and future plans.