Uploading rules to a LOFT section

Linear-on-the-fly testing (LOFT) lets you create a unique test for each candidate using items based on domain and psychometric data. This style of testing ensures item exposure is optimised across your bank of items, meaning that individual items are never overexposed or underutilised.

You can upload an XML file to define the rules for an external optimiser section in a LOFT test.

This article explains how to upload rules for an external optimiser section, including how to create an XML file containing data about your rules.

1. Go to the LOFT/Adaptive screen

To view the LOFT/Adaptive screen, go to Test Creation > LOFT/Adaptive.

TIP: For more information about LOFT/Adaptive, read About the LOFT/Adaptive screen.

2. Select Edit Test Form

To view the Edit Test Form screen, choose a test form in the LOFT/Adaptive grid and select Edit Test Form.

NOTE: You can only edit external optimiser sections in the Edit Test Form screen. For information on how to edit other section types, read About test form content.

The Edit Test Form screen displays information about the chosen test form, including the Test Form Name, Test Form Reference, and any external optimiser sections.

3. Go to the Upload Rules screen

Select Upload Rules in the Edit Test Form screen to upload your rules.

4. Define rules

Choose an XML file and select Open.

Select Download Rules to download your rules. Select Remove to delete your rules.

Creating an XML file

<?xml version="1.0" encoding="utf-8"?>
<rules thetaCut="">
    <content numitems="12">
        <val max="4" min="4" attribute="Animal" type="match">Cow</val>
        <val max="4" min="4" attribute="Animal" type="match">Pig</val>
        <val max="4" min="4" attribute="Animal" type="match">Sheep</val>
        <val max="6" min="6" attribute="Topic" type="match">Diagnostics</val>
        <val max="6" min="6" attribute="Topic" type="match">Treatment</val>
        <val max="4" min="4" attribute="Difficulty" type="match">Easy</val>
        <val max="4" min="4" attribute="Difficulty" type="match">Hard</val>
        <val max="4" min="4" attribute="Difficulty" type="match">Medium</val>
    </content>
    <order>
        <val operator="OR" orderBy="Item Name" order="asc">
            <val type="startsWith" attribute="Question Type" >Information</val>
            <val type="endsWith" attribute="Question Type">Answer</val>
            <val attribute="" type="everything Else" />
        </order>
        <asdf/>
    </rules>

IMPORTANT: Complex ordering, or running large numbers of simulations containing ordering rules could impact performance.

You can create an XML file containing the content rules, psychometric targets (TIF) and item order that the linear optimiser uses to generate test forms. For more information on defining rules and psychometric targets, read Defining rules for a LOFT section and Uploading psychometric targets to a LOFT section.

IMPORTANT: Fields marked with an asterisk (*) are mandatory.

NODE	DESCRIPTION
content*	Collection of attributes to define the content rules for the section.
content / @numItems*	Define the total number of items that should be selected for the section.
content / @useAsOrder	Define the order of items according to the content node. NOTE: Order nodes are no longer required if using the @useAsOrder attribute.
content / val*	The date that the item pool/rules was last edited (based on the server time).
content / val / @attribute*	Specify the name of the item attribute you want to select items from. For example, the name of a tag group.
content / val / @min*	Define the minimum number of items that should be selected to meet the rule.
content / val / @max*	Define the maximum number of items that should be selected to meet the rule.
content / val / @type*	Define the matching type as startsWith, endsWith, containsCaseSensitive, containsCaseInsensitive, sum, range, or match. This defaults to match. NOTE: The sum and range functions must be used with numeric tag values.
content / val / @operator*	Define the boolean operator used to aggregate child nodes. Supported values ‘AND’ / ‘OR’ must be in upper case.
content / val / @minVal	Defines the minimum numeric tag value if the @type attribute is a range.
content / val / @maxVal	Defines the maximum numeric tag value if the @type attribute is a range.
content / val / text ()*	Specify the value you want to use to select items. For example, the name of the tag value.
tifTarget	An optional collection of attributes defining the psychometric targets.
tifTarget / @maxDiff	Define the tolerance for the TIF of all forms generated.
tifTarget / val	Plot individual points for the specified TIF.
tifTarget / val / @theta	Define a theta value on the plotted TIF.
tifTarget / val / text ()	Define a corresponding information value on the TIF.
content / random	Select a random collection of value nodes.
content / random / collection	Define a collection of value nodes to determine the order of items. IMPORTANT: Collection nodes require at least one value node.
order	An optional collection of attributes defining the order of items in the section.
order / val	Define the position for groups of items.
order / @randomiseOrder	Define the order of items selected by each val node as random. NOTE: Individual val nodes can override the @randomiseOrder attribute.
order / val / @orderBy	Specify the name of the item attribute you want to use to order items. For example, the name of a tag group.
order / val / @order	Define whether the items selected using the @orderBy attribute are in ascending (asc) or descending (desc) order.
order / val / @operator	Define the boolean operator used to aggregate child nodes. Supported values ‘AND’ / ‘OR’ must be in upper case.
order/@splitSets	SplitSets indicates that items in an item set will be separated for delivery.
order/@groupSets	GroupSets auto-groups item sets to appear together in delivery. IMPORTANT: GroupSets cannot be used with SplitSets. To use GroupSets, ensure SplitSets is not active.
order/section/distribution	Distribution determines a minimum and maximum number of items into each section. This randomly selects the required number of items from those that have already been selected by the Linear Optimiser. IMPORTANT: Distribution will be invalid if there are not enough items to achieve the minimum. Distribution node is not required for the last section node. This is because it will use the remaining items after distributing to the previous sections.
order/val/restriction	The restriction node ensures that the order is randomised for that val node regardless of the @randomiseOrder value. You can determine where the items and the @topic will appear in the exam
order/section/val/restriction	The restriction nodes within a section are randomised after they have been selected for the section.

5. Save your rules

Select Save in the Edit Test Form screen to save your rules.

You can simulate your LOFT test form to check that the item pool and rules behave as expected for the section. For more information, read Simulating a LOFT test form.

IMPORTANT: You must have defined your item pool before you can simulate your LOFT test form. For more information, read Creating an item pool.