Considering Making Digital Sociology Tools

First off, I believe that websites that help you think are tools. Websites that help you author media are also tools. So when I say “tools” that’s what I mean. I will also likely call them sites. That’s also what I mean. I like to start off posts like these establishing the fact that I’m not going to try very hard to make sense to you but at least I’m explicit about it.

I’ve had a couple of conversations recently that have me close to building these with or without a faculty/course attached.

#metoo Movies

This started when I was watching one of the original Star Wars movies with my kids. I’d never really watched them closely as an adult. It really seemed like Han Solo was a textbook example of a person acting out some pretty specific examples of sexual harassment. Repeatedly Leia tells him to stop touching her and he ignores her etc. I ended up talking to some friends at a school function about it and the conversation expanded into the idea of a database of clips like this that both act as a lens on that time/space/movie and as examples of particular patterns that might be of interest more broadly.

For example here’s Han being creepy . . .

In order to write this post, I’ve already had to write a plugin to handle YouTube snippets better. I built it off this post by labnol.org. Now I can do something like the shortcode below which is handy.
[yt_video id=”xxxxxxxxx” start=”101″ end=”117″]

This contributes to the slow pace of my blog posts.

Anyway, what I’d like to build is a way to display and categorize these particular scenes.

Display could be done through video clips or through animated gifs with subtitles. Giving some simple options to clip out portions like I’ve done above would be nice but would need to support Vimeo and a few others.

Metadata becomes more interesting. The clip excised above would be tied to a few things, some I could automate through a movie database API but others would need human association.

I could see pulling the movie title from some auto-complete field and then tying that to release date, movie studio, maybe a few other things.

The second portion would be about categorizing the pattern of the various issues. There are likely names for these things. For instance Han first ignores two explicit verbal requests from Leia to let her go. Then he follows with a creepy inneundo.

Once we have all this stuff we could display timelines, see incidents per movie, per incident type, per incident type across time etc.

Riots vs Racism

The other tool would be about analyzing news coverage of things like the Eagles super bowl celebration against news coverage of other events like Black Lives Matter protests. This is based on a conversation I had with my wife this morning.

I think this one would be more difficult but really interesting. I’m not sure what would/should be possible mechanically (sentiment analysis? voyant? other stuff?) vs what would be done by hand.

In a perfect world, comparing stories from the same source with the same patterns would be most beneficial. Lining up the facts – simple stuff like geographic location, estimated people in attendance, event cause, etc. The next level would be the linguistic analysis. I think we could use a dictionary API to look for adjectives and/or verbs- mess with various things like that.

2 thoughts on “Considering Making Digital Sociology Tools

  1. Both the database and YouTube clips plugin look like great tools!

    Are you looking at crowdsourcing the clip database? Or curating? (Curious about the frontend you’re imagining.)

    Also, could you use something like the Gifs.com API (https://github.com/gifs/api) to automate the YouTube clips-to-gif conversion + embedding process?

    Keep up the awesome work! 🙂

    1. Right . . . some sort of front-end form submission. I like the Gifs.com API. I had been thinking the ffmpeg model but would prefer an API. My only hesitation on the video-to-gif pattern is the need to have subtitles in a number of cases. The few I’ve looked at on YouTube lack decent subtitles and the auto-generated versions were pretty poor.

Comments are closed.