Assignment Analysis

This document records notes on work of the Measuring qualification effects project, and Refining a framework project, to apply the classification process to student assignments.

Workflow

assignment analysis workflow

The workflow begins with manually downloading the TMAs and EMAs as a zip file. I downloaded them into a folder called zipped_assignments, which is the default, but you can specify a different folder in the scripts.

extract_assignments script

This is specified in the source_folder parameter to the extract_assignments script.

zipped-assignments folder

These are manually prepared zip files containing a presentation’s EMAs and TMAs. By default the script uses a zipped_assignments folder,

assignments folder

If this folder does not exist, the extract_assignments script creates it. You can specify a different folder in the script.

The assignment zip files are unzipped into this folder.

After unzipping there is one folder for each student. Each folder has a sub-folders for each assignment. This sub-folder will contain all of the student submitted files. This means there may be further sub-folders if the student has structured their submission that way, for example, to include supporting data.

extract_text_from scripts

Two scripts extract raw text files from the associated document type. 1 Both scripts write the text output into the source sub-folder.

Other document types can be supported too if the appropriate script is written.

classify_assignments script

Applies the selected classification scheme to the extracted text files.

The individual classification results for each document are written back into the text file’s sub-folder.

The presentation’s summary results are written to the root assignments folder in two formats, json and tsv.

taxonomies folder

The classifiers are set using the standard workflow.

See prepare taxonomy documentation for more information.


1 Text Extraction Methods
  • docx access the file directly to unzip and process the XML. No third party software is required.
  • doc and rtf use OpenOffice to perform the extraction, because it is proven, cross-platform tool software.