Journals are important... so why not make things easier for yourself?
If you really can't get yourself to focus on planning out your day or week, why not do a mind dump?
Essentially, there are two easy steps:
-
Start writing or typing what's on your mind.
-
That's it.
After a pretty big mind dump, there might be some handy stuff in there that you could recycle for other purposes like figuring out what makes us tick.
Sometimes you can get a lot of it yourself, but wouldn't it be nice to have a handy-dandy tool to help you as well?
mindDump's that tool!
Other potential use-cases for this tool include clinical settings that involve social workers and therapists and their respective customers/patients.
There are three separable, mostly independent components to this project that are meant to be modified by the user.
These are:
When originaly setting out to build the mobile app, my top two desires were:
- No spell-check so that way somebody who's dumping their mind won't be bothered to correct their typos.
- Bare minimum aesthetics so that way (once again) there's no distractions when typing down your thoughts.
I used syncthing, an open-source, central serverless continuous file synchronization program that allows you to sync two or more devices' folders together without having to use any cloud services.
ONNX models are known for their interoperability. Something that is desirable if you want to share things with others.
Yes. The model has been fine-tuned on the Go Emotions dataset. After quite a bit of user testing, it appears that there may be some mislabeling in the training data among other issues. These then affect the model's performance on our parsed journal entries.
Yes. Since we're are working with mind-dumped data, punctuation is probably not going to be the best. To account for this, excessive spacing as well as traditional punctuation markers are used to produce sentences.
Yes. The respective column names/format for the csv should consist of doc_id, date, year, month, day, hour, sentence, text, and any number of emotion columns with their respective scores ranging from 0 to 1.
By using the shinylive package and Github Pages. This guide is pretty useful.