betwixt

Betwixt is a lightweight framework for representing semantic assertions as scoped claims and rendering them for human review. It provides a pragmatic review layer between observations and semantic objects, allowing candidate claims to be stabilised through reproducible review workflows.

Betwixt defines a portable tidy claim schema. Implementations reuse their native tabular environments, such as tibble, pandas, or SQLite, while review is rendered through standard HTML/CSS and returned as tidy CSV.

Installation

You can install the development version of betwixt from GitHub with:

# install.packages("pak")
pak::pak("dataobservatory-eu/betwixt")

Example

This is a basic example which shows you how to solve a common problem:

library(betwixt)
ad_gdp_2024 <- claim(
  scope = "country=AD;year=2023",
  subject = "country",
  predicate = "GDP",
  value = "3.73 billion EUR"
)

print(ad_gdp_2024)
#> <claim_df>
#> Claims: 1
#> Scopes: 1
#> Scope: country=AD;year=2023
#> 
#> # A tibble: 1 × 4
#>   scope                subject predicate value           
#>   <chr>                <chr>   <chr>     <chr>           
#> 1 country=AD;year=2023 country GDP       3.73 billion EUR

What is special about using README.Rmd instead of just README.md? You can include R chunks like so:

betwixt_render(ad_gdp_2024)
#> [1] "Claim: country GDP 3.73 billion EUR within country=AD;year=2023"

Vignettes

Betwixt implemeents the following workflow:

observation provenance
         ↓
candidate claim
         ↓
review provenance
         ↓
reviewed claim

It is organised around three complementary vignettes.

1. Preparing Data for Human Review

Introduces scoped claims, semantic stabilisation, contextual inheritance, and the conceptual foundations of Betwixt.

2. R Reference Implementation

Shows how claim_df objects are created, manipulated, rendered, and integrated with the R ecosystem.

3. Python Minimal Implementation

Demonstrates that the Betwixt model is implementation-independent and can be realised using pandas and Mustache-compatible tooling.