General notes

[evamaxfield]

We can try to separate out a few datasets as strictly educational from potentially prototype / in production (iris and boston housing are typically educational, megaface is typically in production)

Use Student / Academic account labels for GitHub accounts as the scraping basis for detecting dataset use.

Add GitHub username to demographic info as a method to get github info for scraping

Describe in three paragraphs:

• Trying to recruit GitHub profiles via survey
• Mine GitHub for use of datasets agnostic of inclusion criteria (as a method for comparing our subset to the larger development community)
• Creating a conceptual model of controversy via encoding and generalizing the datasets on exposing.ai and their reasoning.

[evamaxfield]

For DSSG

What do we imagine the students would be doing? What data would be used, and what are steps for getting / processing the data? Who will be the lead for doing this through the summer?

[evamaxfield]

Roadmap:

• Abstract, higher level, write goals of the project
• Find datasets that meet some criteria that we could potentially scrape GitHub for
• Come up with DSSG proposal
• Come up with survey proposal w/ lab and collection process
• Build scraper

[evamaxfield]

understand and know research design for surveys doing a lit review of "who else has studied controversial datasets, what did they find?" doing the conceptual work ourselves -- where are the gaps -- here is how we are going to validate or test

https://www.zotero.org/groups/4508576/bits/collections/VRK3HG8Z

[evamaxfield]

Download zotaro beta build and take notes on the readings in the app

[evamaxfield]

General notes on literature reviews: http://www.raulpacheco.org/resources/literature-reviews/

"Use the literature review as a method for argument for why our research should be done" -- "what is the coverage and what is the gap" -- "no one has asked these questions"

---

Another approach is finding literature that is similar to what you have done and comparing the results. Common in CS because you may be comparing algorithym performance.

---

"Practice for the second chapter of your dissertation"

---

prospective lit review - going out, seeing what exists, how it matches some criteria for relevance (internal) systematic lit review - externalized, specific query terms

---

For this project: we can use a literature review as an argument for the study. there is a gap in the literature. but it will be hard to find comparable studies. -- this is somewhat of identifying an emerging problem.

---

potential paramters for vignette:

• how much
• acquisition method
• data modality
• end product
• who is using it (students, company, non-profit, gov, etc.)

an additive question about who has access: "you said this was unethical usage, what if users agreed to this use"