You are producing your scientific results using a computing environment such as Python and R?

Here is how you can ensure your scientific contributions - the scientific results you publish with your article - are produced machine readable.

Protip: Don't bury scientific information in text [1] and wait for artificial intelligence to extract it. It won't. Not today, not tomorrow, not with the richness and accuracy the approach described here achieves.

The recipe is straightforward:

1. Install the ORKG Python library or ORKG R package, or both
2. Use ORKG Templates in your Python or R scripts to describe your scientific results in a structured manner (Note: Create templates first if none of the existing fit)
3. Submit this supplementary data together with your manuscript to your favorite journal or conference
4. Alternatively: Deposit the supplementary data with your favorite data repository
5. Contact us so that we can take it from here with the publisher.

That's it. When your article is published, ORKG can simply harvest the data describing your scientific contributions and make it available for reuse, including in ORKG services.

Let's look at an example.

You are researching Irises and one of your questions is whether the setosa and virginica species have different petal length. You conduct a t-test in R.

data(iris)

iris <- iris[iris$Species != "versicolor", c("Species", "Petal.Length")]
iris$Species <- lapply(iris$Species, as.character)

x <- iris[iris$Species == "setosa", "Petal.Length"]
y <- iris[iris$Species == "virginica", "Petal.Length"]

tt <- t.test(x, y, paired = FALSE, alternative = "two.sided", var.equal = FALSE)

In your manuscript you might include a boxplot of the data as follows and state that the p-value is less than .001. With such a plot, the data themselves become scientific information published in the article.

In addition, you can describe your t-test as scientific information in a structured manner.

To do so, you use the ORKG Template for t-test in your R script as follows.

library(orkg)

orkg <- ORKG(host="https://orkg.org")
orkg$templates$materialize_template(template_id = "R12002")
tp = orkg$templates$list_templates()

instance <- tp$students_ttest(
  label="Statistically significant hypothesis test with petal length dependent variable on setosa and virginica irises", 
  has_dependent_variable="http://purl.obolibrary.org/obo/TO_0002605", # petal length
  has_specified_input=tuple(iris, "Setosa and virginica petal length dataset"),
  has_specified_output=tp$pvalue("the p-value", 
    tp$scalar_value_specification(as.character(tt$p.value), tt$p.value)
  ),
)
instance$serialize_to_file("myarticle.contribution.1.json", format="json-ld")

The machine readable description of your t-test is serialized to the file named myarticle.contribution.1.json, which you may submit as supplementary data of your manuscript to your favorite journal or conference.

To ensure that machines can discover such supplementary material given an article DOI, we suggest publishers relate the article and its supplementary materials using the is-supplemented-by relation in Crossref DOI metadata according to the following pattern. An example can be found here.

"relation": {
  "is-supplemented-by": [
    {
      "id-type": "uri",
      "id": "https://example.org/myarticle.contribution.1.json",
      "asserted-by": "subject"
    }
  ]
}

With this, given the DOI of the published article, ORKG can simply harvest both article metadata as well as the data describing your scientific contributions, and make this data available for reuse. An article for our research on Irises could look as show in the following figure.

[1] Mons, B. (2005). Which gene did you mean? In BMC Bioinformatics (Vol. 6, Issue 1). Springer Science and Business Media LLC. https://doi.org/10.1186/1471-2105-6-142