TYPEs¶

TYPE is a property of an OBJECT but in practice it is defined as a separate node within the database to allow us to improve the flexibility of the underlying data model. TYPE nodes are defined either as part of the w3C Annotation model (e.g., {type:'TextualBody'} or using other schema, for example {type:'schema:CodeRepository'}.

type	n
schema:Grant	467,000
TextualBody	356,000
schema:Article	159,000
schema:CodeRepository	74,000
schema:DataCatalog	2,300
Dataset	390

Matching Objects by Type¶

Using the TYPE element we can find (for example) a database using:

  MATCH (tdb:TYPE {type:'schema:DataCatalog'})
  MATCH (tdb)<-[:isType]-(n:OBJECT)
  RETURN n
  LIMIT 3

The TYPE defining an object also defines some of the properties that the object may have, although these are not embedded as constraints.

Various TYPEs¶

schema:DataCatalog¶

Objects defined as data catalogs were initially scraped from re3data.org using a script that has been stored on GitHub in a code repository <http://https://github.com/throughput-ec/throughputdb/tree/master/Re3Databases>_.

The object has properties:

• id, which come from the unique identifier (DOI) assigned by re3data.
• name is the official name of the database.
• url is the URL link for the database.
• keywords is an array of terms associated with the database (although see keywords)
• description is the description given to the database.

An object is represented by the cypher statement:

  (ob:OBJECT {id: "10.17616/R3PD38",
              name: "Neotoma Paleoecology Database",
              url: "http://neotomadb.org",
              keywords: ["paleoenvironments", "fossil mammals (FAUNMAP)"],
              description: "Neotoma is a multiproxy paleoecological database that
              covers the Pliocene-Quaternary . . ."})

Individual databases have keywords associated with them. In some cases keywords may be misspelled, or may be either too general ("science") or too specific ("holocene fossils"). To ensure that keywords can be used effectively we want to add them to the graph.

We do this by creating a special node type (KEYWORD), with the constraint that each keyword is unique. Keywords may be related using the :implies relationship.

Each KEYWORD is related to an OBJECT through annotations.

schema:CodeRepository¶

Objects of type schema:CodeRepository were originally scraped from GitHub, with code that is found in the github_scrapers repository.

The objects have the properties:

• id which is the repository assigned ID (a numeric identifier in the case of GitHub)
• name The name of the repository is a combination of the repository owner and the repository name.
• description A user provided description for the repository.
• url The URL for the repository.

  MERGE (ob:OBJECT {id: 1234456,
                    url: "http://github.com/SimonGoring/myrepo",
                    description: "This repo is the bomb of bombs.",
                    name: "SimonGoring/myrepo"})

Code repositories may have keywords assigned, or have keywords defined (topics in the terminology of GitHub). These are linked to KEYWORD objects in the database through an ANNOTATION.

schema:Article¶

A scientific journal article type is a simplification of the CrossRef data schema v0.1.1. The schema for OBJECTs of type schema:Article provides fields for journal, title, author, and URL. The field id is used for the DOI, to keep a standard of unique IDs in the id field.

• id - Unique identifier for grants (assumes NSF grants)
• journal
• title
• author
• url

MERGE (award:OBJECT {"journal": "Risk Analysis",
                     "id": "10.1111/risa.13567",
                     "title": "Protecting From Malware Obfuscation Attacks Through Adversarial Risk Analysis",
                     "url": "https://onlinelibrary.wiley.com/doi/abs/10.1111/risa.13567",
                     "authors": "Redondo, Alberto; Insua, David Ríos"})

schema:Grant¶

The structure of a grant is (currently) based on the NSF Award schema. The code to obtain the NSF awards and add them to the graph is located within the Throughput database nsf_award module.

• AwardID - Unique identifier for grants (assumes NSF grants)
• name
• amount
• ARRAAmount
• AwardInstrument
• description

MERGE (award:OBJECT {AwardID: 1234567,
                     name: "Award name",
                     amount: 12345,
                     ARRAAmount: 12345,
                     AwardInstrument = "Award Thing",
                     description = "The abstract of our grant proposal."})

Constraints and Indices¶

There are no constraints or indices on TYPE nodes.