4. Graph Data

"Lost my shape - trying to act casual!
Can't stop - I might end up in the hospital
I'm changing my shape - I feel like an accident
They're back! - to explain their experience"
- Talking Heads "Crosseyed And Painless" [Album: Remain in Light]

Introduction

This chapter covers the conversion of the catalog data to RDF triples, data validation using SHACL, and upload to the database.

Data model

"Whatever data you manipulate or store, there is always a model (also referred
to as "schema" or "ontology"). Sometimes this model is not explicit, but it
exists nevertheless, at least in the developers heads and is consequently
reflected in the code that is written to manipulate or consume graph data."
 - Sequeda and Lassila (2021)

Before converting the data to RDF, it is useful to create a diagram of the data model based on the draft ontology, competency questions, and data extracted from the files. The model evolves during code development, improving with each iteration. The flexibility of the graph data facilitates this flexible approach to model development.

A diagram like Figure 4.1 helps verify your thinking during the development process. It is also helpful later when writing SPARQL queries.

Data Conversion

There are many approaches and tools that can be used to convert the collected data to RDF triples. For example, mapping languages like R2RML are used to map relational sources to graph databases. This project uses R code for the data conversion, providing a high level of control of the data and structure at the expense of writing and maintaining many lines of R script.

The R code below illustrates a for loop that processes the dataframe, first assigning a FileType class to the file IRI. Next, if a value for file modification date is present in the data, a triple is formed to represent this information. Many more lines of code will follow to fill out the data model with instance data.

for(i in 1:nrow(srcDf))
{
    # File Type
    rdf_add(some_rdf,
      subject      = paste0(FILECAT, "FILE_", srcDf[i,"filePath_h"]),
       predicate   = paste0(RDF, "type"),
       object      = paste0(FILECAT, "FileType", srcDf[i,"fileType"])
     )

    # Date Modified
    if ( ! is.na (srcDf[i,"modifiedDate"]) ) {
      rdf_add(some_rdf,
        subject      = paste0(FILECAT, "FILE_", srcDf[[i,"filePath_h"]]),
        predicate    = paste0(SCHEMA,  "dateModified"),
        object       = srcDf[i,"modifiedDate"],
        objectType   = "literal",
        datatype_uri = paste0(XSD,"dateTime")
      )
    }
.... more lines of code