Hello! This is a draft of instructions I intend to send out to volunteers for the contextual flashcard dataset collection project. I would love your feedback on the instructions before I send them out more widely.
Welcome to the Contextual Flashcard Dataset Project. Thank you for volunteering! By working together, we are assembling a dataset of thousands of in-context flashcards to help evaluate and improve future automatic flashcard generation projects.
To get started, install the Browserflow extension and the flow: https://browserflow.app/shared/1b2b5231-5f96-4f73-87a6-006626fa3831
Your help is requested in creating flashcards to populate the dataset. Please submit between 5 and 200 flashcards that you believe would be useful or interesting to you. The flow above will assist you in submitting flashcards to the dataset.
To submit a flashcard for the dataset:
- Browse the web as normal.
- Select some text you consider important.
- Run the Create Flashcard Entry flow; it will open the Create Flashcard Entry form and pre-populate it for you.
- Make any desired changes to the “Flashcard Front” and “Flashcard Back” fields in the Create Flashcard Entry form.
- Submit the form.
In lieu of using Browserflow, you may also submit the data entry form manually.
How will the data be used?
The dataset will be cleaned. Personal information or offensive language will be removed. The resulting dataset will be made public for anyone to use.
When are submissions open?
We will begin collecting submissions on December 1, 2021, and leave submissions open for 2 months, closing submissions on January 31, 2022.
What flashcards should I submit?
That’s up to you! Still asking? Here’s some guidance:
Submit any flashcards that you would (1) find personally interesting or useful, and (2) only contain public information.
Interesting facts from Wikipedia or articles you read online are a good idea. Private information about close friends is not.