css.php

Data Mining Project – Tessa Maffucci & Min Huh

Data Mining for a fashion vernacular, #SPREZZATURA 

For the data mining project, Tessa and I (we both are in Fashion Studies track) had tried to retrieve some data for a contemporary fashion vernacular, #sprezzatura. Sprezzatura is an Italian term means “a perfect conduct or performance of something (as an artistic endeavor) without apparent effort.” (http://www.merriam-webster.com/dictionary/sprezzatura) In fashion world, this term has been used to describe “effortlessly stylish” people, especially in menswear world. Utilizing API, we wanted to collect the images, the tags (tagged with sprezzatura), the posting dates, the users’ locations, and so on. We picked two social media platforms; Tumblr and Instagram. However, using each platform’s own API console and getting Authentication token and all seemed like a conundrum to us. So I asked my programer friend who works at Comcast a help (He lives in Mountain View, California). He generously walked us through it.

First of all, he strongly suggested us use Apigee website (https://apigee.com), and it’s FREE! It uses your own account for each social media platform for authentication. He said Apigee would be more than good enough for mining simple data like what Tessa and I wanted. I screenshot every step to share it with our classmates. So, here’s how we collected data via Apigee. (We used Google Chrome browser FYI.)

We started with Instagram. > https://apigee.com/console/instagram
1

Choose “OAuth 2”

2

3

4

5

6

7

8

9

Click this URL > https://chrome.google.com/webstore/detail/jsonview/chklaanhfefbnpoihckbnefhakgolnmc?hl=en
10

11

12

Click this URL > http://www.epochconverter.com/
13

new1

new2

new3

new4

Then, we wanted to try Tumblr but he said, for some reason, Apigee didn’t let him retrieve data from Tumblr. But he could find a way to go around and collect the data from Tumblr (a little bit of cheating I guess). Click this URL > https://www.tumblr.com/docs/en/api/v2#tagged-method
14

15

Then, we tried Twitter just for fun.
16

17

18

18-2

19

 

 

We asked him how we could organize the data on JSON. Then, he showed us how to convert JSON to CSV.

20

21

22

Click this URL > http://konklone.io/json/
23

24

We didn’t get to collecting image data. However, at least we learned that you could (manually) download the images by clicking the links on JSON. My friend is a busy guy so we felt bad for “exploiting” him too long for this project. So, unfortunately, this was the end of our journey. I wish we had more time for cleaning the data we collected and creating some visualization on Gephi!

Here’s the link of the converted CSV file

> https://drive.google.com/file/d/0B0yo41A7AItbU1N3YXdIMWVnSjA/view?usp=sharing

8 thoughts on “Data Mining Project – Tessa Maffucci & Min Huh

  1. Renzo Adler

    Fascinating. Do you know if there’s a way this program can track the “Likes” a post gets in tumblr and the location of the posters doing the liking?
    One project I have in mind isn’t tracking a specific tag, but rather tracking the locations of people that like certain content creators.

  2. Minn Hur Post author

    Alright, Renzo, I think I just figured out how to track the likes for a certain post. But unfortunately it does not provide the locations. It does provide the id numbers though. Here’s one of the examples,

    “username”: “figs__”,
    “bio”: “Minimal fashion.\nLiterature. \n6’4”,
    “website”: “http://figs3.tumblr.com”,
    “profile_picture”: “https://igcdn-photos-f-a.akamaihd.net/hphotos-ak-xap1/10665927_1481043782180037_934581162_a.jpg”,
    “full_name”: “Daniel Figueroa”,
    “id”: “31226107”

  3. Mary Catherine Kinniburgh

    As so many others have pointed out, not only is your documentation so wonderful and clear, but the methodology itself is really solid. Your steps are really logical, and thinking about “reproducible research” as a cornerstone of data inquiry, I think your work easily passes this test.

    Such exciting work!

Comments are closed.