Essay Grading Helper

Table of Content

1. Requirements

2. System Pipline

3. Prequesist

   • Download Roster File
   • Download Assignment Files
   • Organize your files

4. Folder Explanation

5. How to use LightSIDE to predict scores

Requirements

• Python 3
  Python 3, and the dependent packages can be found in requirements.txt file

• LighSIDE
  This is an open source text analysis software; you can also download it here

System Pipeline

Prerequisites

In order to use this system, you also need the roster file/gradebook spreadsheet and the assignments from Blackboard downloaded on your local machine.

Download Roster File From Blackboard

Step 1. Go to the Full Grade Center on Blackboard, hover over Work Offline and click Download.

Step 2. Select the column to the assignment number you want to grade, and check the Include Comments for this Column box.

Step 3. Under Options, we recommend to choose Comma as delimiter type and download as a csv file. (as shown below) Depending on the file you download, make sure you choose the appropriate settings when reading this file. (see section...)

Step 4. Click Submit and you should be able to download your roster file.

Download Assignment Files

Step 1. Go to Assignments section on Blackboard, click the downward arrow of the assignment that you wish to download and select Assignment File Download (see picture on the right)

Step 2. Select all the files (check last attempt file under select file section) and click Submit

Setp 3. The assignment files should be downloaded on your local machine as a zip file.

Organize Your files into appropriate folders

Although you can put your files anywhere you like, we recommend to organized the files in a certain way to make the following process easier to follow and keep your files organized.

1. put your roster file under the folder: ./gradebook/<hw_code>/ e.g. assignment1 in spring 2022 semester could be hw1_s22

2. put your essay files under the folder ./essays/<hw_code>/

Folder explenation

classifiers folder

This folder holds all the trained model classifiers in the folder as well as the training data generated by helperscripts in each according folders

essays

Essays folder holds the raw file of the student submission

gradebook

This folder holds the graded assignments

helperscripts

This folder holds the python helperscripts that helps to transform the docx essays into spreadsheets, including parsing docx, manipulating spreadsheets and post process the rank into score and writing comments.

• aes_system.py
  This will be the only program that end user will be executing because all the other scripts are imported to this file. The user just need to change the appropriate parameters and does not need to go deep in other scripts.

• post_process.py
  This script holds the class PostProcessor

• read_docx.py
  This script holds the class DocxReader

• write_into_gradebook.py
  This script holds the functions for manipulating spreadsheets in pandas DataFrame.

How to use LightSIDE to predict scores

Open the LightSIDE program, and

1. Click on "Predict labels" label
2. load the models (they are under the folder classifiers)
3. load the predict templates ()
4. for each model, click predict.
5. save the result to a specified output And you are ready to go to the next step!