Skip to content

Quiz Manager: Question Collection

Jump to bottom
C. George-Adrian edited this page Nov 14, 2020 · 9 revisions

Contents

  1. Introduction
  2. Human Readable (HR) Format
  3. Moodle XML (MXML) to HR Parser
  4. HR to JSON Parser

Introduction

This solution is used for parsing questions exported in the MXML format to a Human Readable format and a JSON file.

There are 2 shell scripts that can be used to generate the HR and JSON outputs.

  1. ./generate-HR.sh uses MXML files as input and generates HR files
  2. ./generate-JSON.sh uses HR files as input and generates JSON files

Both of these scripts have multiple options and their usage will be explained further in this document. The contents of both HR and JSON outputs are determined by the structure.json file which will be explained below.

This file acts like a config file and describes an individual question (JSON Object), it can be modified if a user wishes to add more tags (an example will be given shortly).

The most basic structure.json file containing only the mandatory fields is the following:

{
    "statement":"",
    "tags" : [
    ],
    "answers": [
        {
            "statement":"", 
            "correct":false, 
            "grade":0.0 
        }
    ],
    "correctAnswersNo":0
}

The tags object is a list that will contain all other tags that have been added by running the ./generate-HR.sh script and are not specified in the structure.json file.

In the JSON output file, tags has the following format:

"tags": [
            {
                "key": "ExtraTag1_Name",
                "values": [
                    "Tag_Value1",
                    "Tag_Value2"
                ]
            }
]

For the rest of the examples in this document, we will be using the following file:

{
    "createdOn":"",
    "lastUsed":"",
    "difficulty":0,
    
    "statement":"",
    "tags" : [
    ],
    "answers": [
        {
            "statement":"", 
            "correct":false, 
            "grade":0.0 
        }
    ],
    "correctAnswersNo":0
}

Human Readable (HR) Format

tag1:value1;tag2:value2;...;tagN:valueN;
Statement
+ Correct Answer
- Wrong Answer

Line by line explanations of the above format

tag1:value1;tag2:value2;...;tagN:valueN;

The first line of a question is the tagline. Here all tags and their values are stored. The type of each tag is known from the structure.json file.

All tags that are not found in the structure.json file will be added to the tags field, and all values will be added to a list of strings.

Statement

This is the statement of a question and it will be added as a string to the "statement" field.

Multiline statements are supported.

+ Correct Answer

All correct answers are marked with a '+' character on the first position of the line.

When a correct answer is found "correctAnswersNo" is incremented by 1 and the answer's "correct" field is set to true

After all answers have been read, each correct answers' "grade" == 1/"correctAnswersNo"

- Wrong Answer

All wrong answers are marked with a '-' character on the first position of the line.

For each wrong answer "grade" == -0.5

Multiline answers are supported

Moodle XML (MXML) to HR Parser

More information about ./generate-HR.sh options can be found here

As mentioned above ./generate-HR.sh parses MXML files to the HR format described above.

For this demo we have the following input file exported from Moodle:

MXML

<?xml version="1.0" encoding="UTF-8"?>
<quiz>
<!-- question: DEMO_QUESTION  -->
  <question type="multichoice">
    <name>
      <text>Test Question</text>
    </name>
    <questiontext format="html">
      <text>Test Question Statement?</text>
    </questiontext>
    <generalfeedback format="html">
      <text></text>
    </generalfeedback>
    <defaultgrade>1.0000000</defaultgrade>
    <penalty>0.3333333</penalty>
    <hidden>0</hidden>
    <idnumber></idnumber>
    <single>true</single>
    <shuffleanswers>true</shuffleanswers>
    <answernumbering>abc</answernumbering>
    <correctfeedback format="html">
      <text></text>
    </correctfeedback>
    <partiallycorrectfeedback format="html">
      <text></text>
    </partiallycorrectfeedback>
    <incorrectfeedback format="html">
      <text></text>
    </incorrectfeedback>
    <answer fraction="100" format="html">
      <text>Answer 1</text>
      <feedback format="html">
        <text></text>
      </feedback>
    </answer>
    <answer fraction="-50" format="html">
      <text>Answer 2</text>
      <feedback format="html">
        <text></text>
      </feedback>
    </answer>
    <answer fraction="-50" format="html">
      <text>Answer 3</text>
      <feedback format="html">
        <text></text>
      </feedback>
    </answer>
    <answer fraction="100" format="html">
      <text>Answer 4</text>
      <feedback format="html">
        <text></text>
      </feedback>
    </answer>
    <answer fraction="100" format="html">
      <text>Answer 5</text>
      <feedback format="html">
        <text></text>
      </feedback>
    </answer>
  </question>
</quiz>

Running the following script:

./generate-HR.sh -e topics,author,reviewed_by -u GC -t -c -y

-e topics,author,reviewed_by

This option will add the following extra tags: topics, author, reviewed_by

-u GC

This will autocomplete the "reviewed_by" tag with "GC"

-t

This will use the topic dictionary to automatically complete the appropriate topics for the question

The autotopic dictionary we will be using for this demo is:

topicDict = {}

topicDict["Topic1"] = ["Test", "Dummy"]
topicDict["Topic2"] = ["Question"]
topicDict["Topic3"] = ["Florin", "Salam"]

topicList = list(topicDict.keys())

The autotopic component of the Quiz Collection solution works by matching individual words from the question's statement and its correct answers to words related to a topic. In this demo the word "Florin" and "Salam" are related to the "Topic3" topic. In a real application we could see the words "passwd", "chmod" and "chown" relating to the "users" topic.

For a future feature, a way to generate this topic dictionary would be to feed relevant pieces of text to a script and generate a word cloud, with the most frequent terms being used for the dictionary.

! Warning

HR to JSON Parser

More information about ./generate-JSON.sh options can be found here

Clone this wiki locally