Reflections and considerations thus far
Thus far, I have undertaken some interesting work looking at the networks between different Fridays4Future accounts. This work is more in the field of testing the veracity of different solutions and the limits of my software and hardware, with the aim of informing a more detailed research project at some point in the future. There is still the outstanding issue of ethics, as well as some theoretical work to be done.
In short, I have identified and tested a method by which I can collate followers of certain accounts and then indicate which followers they all have in common via a social network analysis graph. This works using InstaScrapr (a github python program), R (an open source statistical analysis programming language) and Giphy (an open source visualisation tool). The end result is a plot that shows a range of nodes (organised into coloured clusters) with links between the nodes (and clusters).
This might go some way to exploring the density of a network, as well as the centrality of some of the nodes within that network. This might be useful for considering the ways in which learning takes place across social media networks. I guess we’re not looking so much at what is being learnt, but the mechanisms by which learning happens – the channels, to use some very old fashioned communications theory language. Fine.
Some weaknesses have caused frustration. Principally, at this stage, my network analysis doesn’t seem to allow me to geographically locate the accounts. This is a problem that I can foresee because I can’t scrape the accounts from, for example, Sydney, or NSW, and then look at how they’re working together. Of course, I could use some search tools to identify Australian Instagram accounts – but there is always the risk of missing something there. I think this is probably going to be more of a problem when I start looking at specific hashtags, and not accounts – so I will note this problem and then look to work out a solution at a later date.
Q: Can I filter so we only look at the accounts that use a geotag? Is that possible?
Another issue appears to relate to the fact that I am only scraping first order followers; that is, I am not scraping the followers of followers. There’s a good reason for that – it would quickly move beyond the realm of feasibility if that was the case. However, I do wonder if this limits the validity of my data. For example, if Person A and Person B both follow Person C (but they don’t follow each other), there’s a kind of networked connection there – but unless i have scraped Person C, I wouldn’t know of that connection. I think this is often referred to as a mutual.
Q: What can be done about mutual connections?
Ethical concerns.
Yeah. I have them.