Gradeer is a modular assessment tool with support for both automated and manual assessment.

Gradeer is compatible with JDK 1.8. We have also tested it with JDK 11; later versions should be compatible.

Apache Maven is required to build and test Gradeer.
Maven can be installed by sudo apt install maven on Ubuntu.

In order to use PMD checks, PMD must be installed. We officially support version 6.49.0, but newer versions may also work.

You must also set the environment variable GRADEER_PMD_LOCATION to point to PMD's directory (which contains the bin and lib directories), e.g. /home/username/Software/pmd-bin-6.24.0/

Clone and enter the repository.

git clone https://github.com/ben-clegg/gradeer && cd gradeer

Run the build process with Maven. This will also run some unit tests; if any tests fail, check that the prerequisites are correctly installed and configured, and repeat the process.

mvn clean package

Tests can be run alone with the following command:

mvn test

Gradeer is executed by:

java -jar <path to gradeer jar with dependencies> -c <location of configuration JSON file>

The --help command displays other available execution options.

java -jar <path to gradeer jar with dependencies> --help

This repository provides an example environment, called "liftPackaged." While standalone Gradeer binaries are available, we recommend building the tool from source (as above) so a copy of this enviroment is available. This example environment can be executed from the root of the repository by:

java -jar target/gradeer-0.98b-jar-with-dependencies.jar -c src/test/resources/testEnvironments/liftPackaged/gconfig-liftpackaged.json

Gradeer performs the following in the execution environment:

• Load checks, model solution, and "students'" solutions.
• Compile model and "students'" solutions.
• Run checks on model solution to identify and remove faulty checks (UnitTestCheck_TestDummyB and UnitTestCheck_TestDummyC are faulty). These are reported in the environment's output directory.
• Run remaining checks on "students'" solutions.
• Store grades and feedback in the environment's output directory.

When making a grading environment for Gradeer, it should be contained within its own directory. This should be structured similarly to the example environment, present in src/test/resources/testEnvironments/liftPackaged.

The main configuration file is a JSON file, which defines Gradeer's execution parameters. Any path-based parameters can be defined in an absolute manner, or (preferably) relatively to this JSON file. The following is an explanation of the example environment's configuration, gconfig-liftpackaged.json:

{
  // Location of students' solutions
  "studentSolutionsDirPath": "studentSolutions",
  // Location of known correct model solution(s)
  "modelSolutionsDirPath": "modelSolutions",
  // Location of test suites
  "testsDirPath": "testSuites",
  // Locations of check definition JSON files. Files can contain different types of checks, but it may be beneficial to group them.
  "checkJSONs": ["unittestChecks.json", "checkstyleChecks.json", "pmdChecks.json"],
  // Time budget for each test suite to execute. Tests are assumed to fail otherwise.
  "perTestSuiteTimeout": 120,
  // Location of checkstyle rule definitions
  "checkstyleXml": "checkstyle.xml"
}

Checks can be defined across multiple JSON files, each containing an array of one or more check definitions. Each check requires a name and a check type. Other parameters can also be defined, such as weights and feedback. For example, the following would define a check for each currently implemented type of check:

[
  {
    "type": "CheckstyleCheck",
    "name": "MethodName",
    "feedbackValues": [
      {"score": 0.0, "feedback":  "One or more method names not in camelCase."},
      {"score": 1.0, "feedback":  "Method names use correct camelCase."}
    ]
  },
  {
    "type": "PMDCheck",
    "name": "EmptyIfStmt",
    "weight": 0.5,
    "feedbackValues": [
      {"score": 1.0, "feedback": ""},
      {"score": 0.0, "feedback": "Your code contains empty if statements. This should be avoided as it reduces readability."}
    ],
    "maxViolations": 2
  },
  {
    "type": "UnitTestCheck",
    "name": "TestLiftA",
    "weight": 8.0,
    "feedbackValues": [
      {"score": 0.0, "feedback":  "TestLiftA Incorrect"},
      {"score": 1.0, "feedback":  "TestLiftA Correct"}
    ]
  },
  {
    "type": "ManualCheck",
    "name": "Error Reporting",
    "prompt": "Quality of error reporting?",
    "weight": 4.0,
    "maxRange": 10,
    "feedbackValues": [
      {"score": 0.9, "feedback": "Excellent error reporting"},
      {"score": 0.7, "feedback": "Good error reporting"},
      {"score": 0.5, "feedback": "Average error reporting"},
      {"score": 0.0, "feedback": "Poor error reporting"}
    ]
  }
]

When executed, each check calculates a normalized score between 0 (complete failure) and 1 (perfect pass). These scores are used to determine which band of feedback to provide, as defined by "feedbackValues".

When calculating the final grade of a solution, the score of each check is multiplied by its weight, then the sum of these values is divided by the sum of checks' weights.

Development of this tool was funded in part by The University of Sheffield.