Covid narratives

Step 0:

• Follow the Getting Started instructions to install CRISP-T.

Step 1: Download data

• Download COVID narratives data to crisp_source folder in home directory or current directory using crisp --covid covidstories.omeka.net --source crisp_source.
• Download Psychological Effects of COVID dataset to crisp_source folder.

Step 2: Data Import

• Create a crisp_input folder in the current directory for keeping imported data for analysis. Import data from crisp_source folder to crisp_input folder with:

crisp --source crisp_source --out crisp_input

• Ignore warnings related to PDF files.

Step 3: Qualitative analysis

• Perform topic modelling and assign topics(keywords) to each narrative, redirecting the output to the same folder.

crisp --inp crisp_input --assign --out crisp_input

• Explore results

crisp --inp crisp_input --print "documents 10"
crisp --inp crisp_input --print "documents metadata"

• You will notice interviewee and interviewer keywords. These are assigned based on the presence of these words in the narratives and may not be useful.
• You may remove these keywords by using --ignore with assign and recheck the results.
• --clear option clears the cache before running the analysis.
• ⚠️ While analyzing multiple datasets, use crisp --clear option to clear the cache before switching datasets. ⚠️

crisp --inp crisp_input --clear --assign --ignore interviewee,interviewer --out crisp_input
crisp --inp crisp_input --print "documents metadata"

• Prints the clusters assigned to each document based on keywords.

crisp --inp crisp_input --print "metadata clusters"

• There are many other options to explore the results. See documentation for details.
• Let us choose narratives that contain 'mask' keyword and show the concepts/topics in these narratives.

crisp --inp crisp_input --clear --filters keywords=mask --topics

• The above results will not be saved as --out is not specified.
• Notice time, people as topics in this subset of narratives.
• When using filters, you should explicitly specify --inp and --out options with different folders to avoid overwriting the input data.