[QTI] IMS QTI V2.1 PD (revision 2)
Yvonne Winkelmann
YW0104 at gmx.de
Fri Oct 20 15:40:26 BST 2006
Hello Steve Lay,
thanks for your responses.
> > our corporation develops a test modul for established lms like OLAT
> > adapted from IMS QTI version 2.1 public draft.
> This sounds exciting, will your code be part of OLAT itself?
We are developing test logic and player components for a special project regarding language testing. These QTI v2.1 based components will form an integrated solution together with OLAT, so a basic QTI v2.1 implementation will be part of a new OLAT release in 2007. Separately we create a standalone java based QTI v2.1 authoring tool in cooperation with another LMS provider, Metacoon.
I read your article about reports and it is really helpful but I still have questions for the new result file description and some qti example files.
--
The result specification labels the item variables numAttempts, duration and completionStatus as "build-in variables". I understand this declaration in the way that these variables are not declared in the items and assessments itself but shall be stored by the system for these resources.
In your article you explain that the responseVariables numAttempts and duration can be used in the responseProcessing of the resources. This is really useful but the QTI examples show this aspect in a confusing way.
Example: "items/adaptive.xml"
This example sets the outcomeValue "completion_status". This variable is not declared by responseDeclaration of the item. I do not know from where the system should know this variable. From the systems point of view I see a problem because the variable completionStatus is not a required build-in variable and so the system tries to set a variable which is not declared (apart from the spelling mistake - I hope numAttempts, duration and completionStatus are identifiers).
Example: "items/feedback_adaptive.xml"
This example declares the responseVariable "ATTEMPTS" which is used in the responseProcessing. I do not understand why the build-in variable "numAttempts" was not used in this context.
Regrettably none of the QTI examples shows the use of the build-in variables in an expedient way.
--
I see a problem in the calculation of how long a learner spent time on an item (duration). In our tool we want to offer an itembank so you can choose some existing items and compile a test. I think the items should be independent from the test.
For a testPart we want to support the submission modes "simultaneous" and "individual". If an author declares a responseProcessing where he uses the build-in variable "duration" and the item was used in a test with "simultaneous" mode it is an error. No system can declare duration or timeLimits for items in "simultaneous" mode. A timeLimit can be ignored by the system if there is not another possibility but what happens if the duration variable is used for the responseProcessing?
In this context we have also another "problem":
Currently we present all items per section to one screen page and use the duration variable for learner tracking. So the evaluator is able to reproduce how long the learner saw the items on this page. The duration was saved per item. This means that all items per section have the same duration and the sum of all items is not the test duration. That seems wrong. But if you save the duration as responseVariable of the test relating to a section one does not know by viewing the assessmentResult only which items belong to which section because the structure of the test is not saved in the result.
(In my opinion the storage of the correctResponse in the assessmentResult since QTI v2.1 revision 2 shows that some information needs to be saved doubled.)
--
Sorry for the many and long questions but I take the liberty to ask one more question concerning the storage of the assessmentResult. In my opinion one content package may contain more than one test. But one assessmentResult contains only one test which means that I have to create several result files per content package. I believe I have to save the relation between test and result in the manifest. How the specification envisage this?
--
At least I have seen some noticeable problems in the examples:
-
example: "tests/report.xml"
1. schemaLocation="http://www.imsglobal.org/xsd/imsqti_v2p1 http://www.imsglobal.org/xsd/imsqti_v2p1.xsd"
The website represent the old schema and so the xml file does not validate against the schema.
2. responseVariable identifier="numAttempts"
How can the value of the reponseVariable "numAttempts" for two itemResults of the item Q01 be "1"?
-
example: "items/choice_multiple_chocolade.xml"
<match>
<variable identifier="MR01"/>
<baseValue baseType="identifier">C01 C02 C03 C04 C05 C06 C07 C08 C09 C10</baseValue>
</match>
The responseVariable "MR01" is a multipleValue. Matching the responseVariable "MR01" and the declared singleValue is a system side problem. The values never are the same.
>From my point of view the matching should happen in the same way the example "items/order_partial_scoring.xml" shows.
Thanks in advance
Yvonne
--
Yvonne Winkelmann
Dipl. Medien-Informatikerin
BPS Bildungsportal Sachsen GmbH
Bahnhofstraße 6
09111 Chemnitz
web: http://www.bps-system.de
----------------------------------
--
Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer
More information about the IMS-QTI
mailing list