Code to automatically extract data from Autobiographical Interview (AI) transcripts, scored by hand in Microsoft Word.
Robert Seymour. (2022). neurofractal/autoAIscore: v0.11 (v0.11). Zenodo. https://doi.org/10.5281/zenodo.7225874
- Download and Install Anaconda Python from HERE
- Download the autoAIscore repository in Documents (or wherever you want it)
- Open Anaconda3 --> Anaconda Prompt
- In the terminal type:
pip install python-docx lxml
- In the terminal type:
# Change path to match your Documents path
cd /users/XXX/Documents/autoAIscore
- To test whether the script(s) are working, in the terminal type:
# Change paths to match your data
python process_AIscores /users/XXX/Documents/autoAIscore/test_AI /users/XXX/Documents/autoAIscore/test_AI/out.csv
!! Email Rob or IT Support if you have any issues !!
- One document should hold transcriptions each participant and for all events
- Events should be clearly separated by 'Event XX' before each transcription of the event
- For each event, split details into new paragraphs (new line) according to AI scoring protocols
- For each detail you wish to score, highlight the relevent text and add a new comment
Speed this up by creating a keyboard shortcut for 'New Comment'
- The comment should contain a 4 letter code, corresponding to the (adapted) Autobiographical Interview scoring protocol:
- Letter 1: I (Internal Detail) or E (External Detail)
- Letter 2-3: EV (Event), PE (Perceptual), TI (Time), PL (Place), TH (Thought/Emotion), SE (Semantic), RE (Repetition), OT (Other)
- Letter 4: T (True), F (False), U (Unverifiable)
It should look like this:
In the command line run (for example):
python3 process_AIscores_filbury.py test_filbury/test_participant01.docx
The script should print any warnings/errors encountered, and export a .csv file with the same name as the .docx file, organised like so:
- Each folder should contain separate transcription documents for each memory, from one participant and NO OTHER .docx FILES
- Split details into new paragraphs (new line) according to AI scoring protocol
- For each detail you wish to score, highlight the relevent text and add a new comment
Speed this up by creating a keyboard shortcut for 'New Comment'
- The comment should contain a 3 letter code, corresponding to the Autobiographical Interview scoring protocol:
- Letter 1: I (Internal Detail) or E (External Detail)
- Letter 2-3: EV (Event), PE (Perceptual), TI (Time), PL (Place), TH (Thought/Emotion), SE (Semantic), RE (Repetition), OT (Other)
It should look like this:
In the command line run (for example):
python3 process_AIscores.py test_AI/ testAI/out.csv
The script should print any warnings/errors encountered, and export a .csv file organised like so: