Pitch 1: Website that creates tangible, engaging visualizations of probabilities.
Humans are poor at conceptualizing probability. I notice that one common reason anti-vaxxers use for refusing the vaccine is that they’d rather take their chances with catching covid than get the shot and come down with complications, which is far rarer than the chance of dying from covid. The objective of this project would be to create a website with high user engagement that would inform users of the odds of certain events happening using tangible, engaging visualizations. The user would be able to select from a list of events to compare. Visualizations that represent a dangerous event that is likely to occur will look visually more dangerous than those that are less likely to occur.
I have found some potentially useful datasets to use
- What kinds of graphics would help reduce the worry that something might happen? Colors, structure, font, etc.
- Do moving images impact the user more than static images? Do they keep users on the site?
- Note: I am having a hard time finding existing sources, and I think my keywords I am using are wrong.
Some sources I found that may be useful/interesting:
- A Literature Review: Website Design and User Engagement (nih.gov)
- Colors and Emotions: Preferences and Combinations: The Journal of General Psychology: Vol 122, No 1 (tandfonline.com)
Pitch 2: Formality indicator for Japanese text
Google translate and Deepl don’t really give users the option to indicate what context a user needs to use Japanese in. One key thing that distinguishes different styles of speech or written communication is the choice in vocabulary. On Jisho, an online Japanese dictionary, some words are marked with what context they are used in. i.e. colloquial, slang, sonkeigo,(Honorific or respectful language). My proposed project would use web scraping from online dictionaries to gather data about which words are used in which context and would tell the user what level of formality the text is in and who it would be appropriate to say/send to.
Ana sent me a list of tools that may be useful. Sudachipy seems promising
- Japan Dict has lists for certain types of words. Sonkeigo is included, as well as gendered language, manga and internet slang, and a lot of other stuff. Tags and lists – JapanDict: Japanese Dictionary
Pitch 3: Visualization of cause of death
I feel like death statistics are inherently somewhat dehumanizing, and I want to use recent death statistics to create a more human representation. The user would be able to enter certain demographics, such as location, age, sex, etc. This would be pretty similar to pitch one, but I would want to represent each person with something that people would empathize with. Perhaps instead of using the data outright, I would gather patterns from the data to create a fake population. I’m not sure how I might do this
- What kinds of images do people empathize with? Would it be better to go with something more amorphous, such as blobs with cute faces or stick figures?
- Keeping things tactful while tackling a sensitive issue.
I have found several sources for cause of death data on kaggle, but not all are completely recent, which is ideally I am looking for. I did find recent data specifically for Brazil. The CDC has some data as well, I just need to figure out how to get at it.