observing-obsidian

Using python, NLP and text vectors to explore my daily notes over the past 8 years.

Designed to work with notes stored in an obsidian vault that are named in the format yyyy-mm-dd.

Get running

• Activate venv source env/scripts/activate
• Install requirements pip install -r requirements.txt
• Run the app py app.py

Example output

Validation of sentiment analysis

Very negative

• 2024-02-12 - was making notes about consequences of product failure (death, injury, harm ect)
• 2016-09-30 - reality was talking about a nerf gun battle - was actually a great day

Very Positive

• 2016-09-04 - marked as highest sentiment but contained only "Dis the oje, went really well", not sure why marked so high
• 2020-11-21 - likely high sentiment due to old method of note taking (dailyo) that included daily tags that marked what i did that day such as #party, #guitar ect

Todo

• parameterise folder location
• exclude dailyo style daily tags, they're interfering with analysis
• make the graph prettier
• vectorise the content of each day
• explore other forms of analysis and look into the approach used by TextBlob