Skip to content

rml-exercise.md

Latest commit

 

History

History
218 lines (158 loc) · 6.75 KB

rml-exercise.md

File metadata and controls

218 lines (158 loc) · 6.75 KB

RML Mapping Exercises

Setup

For these exercises, we'll use Matey - a browser-based tool for creating and testing RML mappings using the YARRRML syntax (a human-friendly notation for RML).

Tool: https://rml.io/yarrrml/matey/#

Goal: Learn how to map CSV data to RDF triples using RML, extending our patient monitoring use case.

Exercise 1: Add Patient Fall Risk Score

Background

We have a CSV file with patient information that's already partially mapped to RDF. Currently, we're mapping the patient ID, room location, and assigned nurse, but we're not yet capturing the fall risk score — a critical piece of information for patient safety.

Current Data (patients.csv)

patient_mrn,patient_name,fall_risk_score,room_location,assigned_nurse_id
MRN-7834521,Maria Van den Berg,85,3-WEST-214,NURSE-102
MRN-7834522,Jan Vermeulen,45,3-WEST-215,NURSE-102
MRN-7834523,Sophie Dubois,92,ICU-3W-216,NURSE-103
MRN-7834524,Lucas Peeters,38,2-EAST-101,NURSE-104
MRN-7834525,Emma Janssens,88,ICU-3W-217,NURSE-103

Current RML Mapping (YARRRML)

prefixes:
  example: http://www.example.com/
mappings:
  patients:
    sources:
      - [patients.csv~csv]
    s: example:patient_$(patient_mrn)
    po:
      - [a, example:Patient]
      - [example:hasID, $(patient_mrn)]
      - [example:locatedIn, example:room_$(room_location)~iri]
      - [example:assignedTo, example:nurse$(assigned_nurse_id)~iri]

Your Task

Add the fall risk score property to the mapping.

  1. Open Matey: https://rml.io/yarrrml/matey/#
  2. Copy the patient CSV data into the "Input: Data Source" section (middle left)
  3. Copy the current YARRRML mapping into the "YARRRML" section (middle right)
  4. Extend the mapping to include a new property: example:hasFallRiskScore
    • The value should come from the fall_risk_score column
  5. Click "Integrate!" to see the generated RDF output

Expected Output (excerpt)

You should see triples like:

example:patient_MRN-7834521 a example:Patient ;
    example:hasID "MRN-7834521" ;
    example:hasFallRiskScore "85" ;
    example:locatedIn example:room_3-WEST-214 ;
    example:assignedTo example:nurseNURSE-102 .

Exercise 2: Add BMS Room Data Source

Background

Our Building Management System (BMS) tracks room conditions and metadata. Currently, our RDF only references rooms by ID (e.g., example:room_3-WEST-214), but we have no information about the rooms themselves — their type, capacity, environmental conditions, etc.

In this exercise, you'll add a second data source and create a second mapping to describe the rooms.

New Data Source (rooms.csv)

Create this as a second CSV file:

room_id,room_type,zone,capacity,temperature_celsius,humidity_percent,hvac_status,last_update_minutes
3-WEST-214,GeneralRoom,3-WEST,2,21.5,72,operational,15
3-WEST-215,GeneralRoom,3-WEST,1,22.0,58,operational,15
ICU-3W-216,ContactIsolationRoom,ICU-3W,1,20.8,55,operational,15
2-EAST-101,GeneralRoom,2-EAST,2,21.2,60,operational,15
ICU-3W-217,AirborneIsolationRoom,ICU-3W,1,21.0,70,operational,15
3-WEST-220,AirborneIsolationRoom,3-WEST,1,20.5,65,maintenance_required,120
ICU-3W-218,GeneralRoom,ICU-3W,1,21.8,62,operational,15

Note: This includes the rooms from our patient data plus a few additional rooms to show different types.

Room Types

  • GeneralRoom: Standard patient room
  • ContactIsolationRoom: For patients with contact-transmitted infections (e.g., MRSA, C. diff)
  • AirborneIsolationRoom: Negative pressure room for airborne infections (e.g., TB, COVID)

Your Task

Create a second mapping for room data.

  1. In Matey, add the rooms.csv data to your input (you can have multiple CSV blocks separated by ---)
  2. Add a new mapping called rooms to your YARRRML file
  3. The mapping should:
    • Use rooms.csv as the source
    • Create subjects with IRI: example:room_$(room_id)
    • Map the room to its type as a class (e.g., example:GeneralRoom)
      • Hint: Use [a, example:$(room_type)~iri] to make the room type a class
    • Include properties for:
      • example:hasZone → zone
      • example:hasCapacity → capacity
      • example:hasTemperature → temperature_celsius
      • example:hasHumidity → humidity_percent
      • example:hasHVACStatus → hvac_status
      • example:lastUpdatedMinutesAgo → last_update_minutes

Starter Template

prefixes:
  example: http://www.example.com/

mappings:
  patients:
    sources:
      - [patients.csv~csv]
    s: example:patient_$(patient_mrn)
    po:
      - [a, example:Patient]
      - [example:hasID, $(patient_mrn)]
      - [example:hasFallRiskScore, $(fall_risk_score)]
      - [example:locatedIn, example:room_$(room_location)~iri]
      - [example:assignedTo, example:nurse$(assigned_nurse_id)~iri]
  
  rooms:
    sources:
      - [rooms.csv~csv]
    s: example:room_$(room_id)
    po:
      # TODO: Add your mappings here
      - [a, example:$(room_type)~iri]
      # Add more properties...

Expected Output (excerpt)

You should see triples like:

example:room_3-WEST-214 a example:GeneralRoom ;
    example:hasZone example:zone_3-WEST ;
    example:hasCapacity "2" ;
    example:hasTemperature "21.5" ;
    example:hasHumidity "72" ;
    example:hasHVACStatus "operational" ;
    example:lastUpdatedMinutesAgo "15" .

example:room_ICU-3W-216 a example:ContactIsolationRoom ;
    example:hasZone example:zone_ICU-3W ;
    example:hasCapacity "1" ;
    example:hasTemperature "20.8" ;
    example:hasHumidity "55" ;
    example:hasHVACStatus "operational" ;
    example:lastUpdatedMinutesAgo "15" .

Bonus Challenge: Add Data Type Specifications

Task

Modify your mappings to specify proper data types for numeric values:

  • fall_risk_score should be xsd:integer
  • capacity should be xsd:integer
  • temperature_celsius should be xsd:decimal
  • humidity_percent should be xsd:integer
  • last_update_minutes should be xsd:integer

Hint: Use the datatype annotation syntax: $(column_name)~xsd:integer

Example:

- [example:hasCapacity, $(capacity), xsd:integer]

Resources

Tips

  • Start simple: Get the basic mapping working before adding complexity
  • Check the output: Always look at the generated RDF to verify it matches your intent
  • Use meaningful IRIs: Make sure your room and patient IRIs match between mappings
  • Test incrementally: Add one property at a time and verify each works
  • Datatypes matter: Numeric data should be typed appropriately for later queries

Good luck!