Description
The second graph shows the main types of images shared and connected with the hashtag #opioid. As the first research has pointed out that the hashtag #opioid is the most popular (the majority of the posts are connected with it) we therefore downloaded 250 posts with more than 50 likes through Instaloader, and that have lately been processed through a tagging process by image recognition using Imagga. Thereafter, the tags with a confidence superior to 30% have been retained and input on Open Refine. The Excel file was than imported in Gephi to obtain a spatial distribution of the images clustered by type of content. After linking the images with Python we cleaned the visualization according to the most relevant tags that were previously found by Imagga.
Focusing on the imaginary of opioids in the U.S. it seemed proper to analyse Instagram as main social network, as its primary domain, as basically everyone knows, is pictures. Given its image-based character, it gives us the occasion to investigate the perception of opioids between the users and, in other words, people.
Protocol
Data
Data Source: DensityDesign
Timestamp: November 2018
View Data (500Kb)
Words by frequency in the scripts of youtube vlogs: the information were obtained scraping the script using “ Tag cloud combinator” and cleaned using Atom.