Quiz Datasets

Quizzes use the evaluation dataset entries (see the schema).

Room Configuration

• Room configurations get a new key, quizzes, with a value which is a mapping defining quizzes which can be run in the room (see rooms/README.md).

API

GET /api/v1/rooms

Room config entries in the response include the value of quizzes (a list of mappings).

GET /api/v1/rooms/{room_id}

Response includes the value of quizzes (a list of mappings).

GET /api/v1/rooms/{room_id}/quiz/{quiz_id}

Fetch the quiz. Returns a mapping for the quiz with a list of questions:

{
    "id": "<quiz_id>",
    "questions": [
        {
            "uuid": "<question_uuid>",
            "inputs": "What color is the sky? (QA)",
            "metadata": {
                "type": "qa"
            }
        },
        {
            "uuid": "<question_uuid>",
            "inputs": "The color of the sky is _____",
            "metadata": {
                "type": "fill-blank"
            }
        },
        {
            "uuid": "<question_uuid>",
            "inputs": "Is the sky blue?",
            "metadata": {
                "type": "multiple-choice",
                "options": {
                    "true",
                    "false"
                }
            }
        },
        {
            "uuid": "<question_uuid>",
            "inputs": "What color is the sky? (MC)",
            "metadata": {
                "type": "multiple-choice",
                "options": {
                    "red",
                    "green",
                    "blue"
                }
            }
        },
    ...
    ]
}

• question-type is one of "qa", "fill-blank", or "multiple-choice"

• Questions of type "multiple-choice" contain an additional entry, "options" in their metadata, whose value is a list of strings.

POST /api/v1/rooms/{room_id}/quiz/{quiz_id}/{question_uuid}

Check an answer. Client should send the answer entered / selected by the user as a text/plain body. For a correct answer, returns:

{
    "correct": "true"
}

For an incorrect answer, returns:

{
    "correct": "false"
    "expected_output": "<expected_output>"
}

(Later, this response might include more information, such as a citation).