Show HN: PotterVerse – Visualizing the Harry Potter character network(efekarakus.com)
efekarakus.com
Show HN: PotterVerse – Visualizing the Harry Potter character network
http://efekarakus.com/potterverse
A visualization of the Harry Potter character network. Exploring whether it exhibits the same properties as a real social network.
8 comments
I'm not sure comparing this to a "real" social network is meaningful. More appropriate might be to compare the Potterverse to another relatively small, socially isolated, insular community like a group of Amish. I wonder how the numbers would compare in that case.
Or even a college which the average distance between any two people is 2 - 3 degrees. Additionally, this story is written from the perspective of Harry Potter so weaker ties won't be included as it doesn't advance the story.
Source: https://www.facebook.com/notes/facebook-data-team/anatomy-of... -- "When we limit our analysis to a single country, be it the US, Sweden, Italy, or any other, we find that the world gets even smaller, and most pairs of people are only separated by 3 degrees"
Source: https://www.facebook.com/notes/facebook-data-team/anatomy-of... -- "When we limit our analysis to a single country, be it the US, Sweden, Italy, or any other, we find that the world gets even smaller, and most pairs of people are only separated by 3 degrees"
That's a great question! I kinda wanted to see if a fictional network would look somewhat similar to a real social network such as Facebook. But exploring other networks that are more secluded would be so interesting! Gathering the data would be a challenge :)
I don't have my copy with me right now, but Barabasi has some examples of interesting social network data that were collected by sociologists in his book "Linked" [1]. You might be able to get hold of some of those. IIRC there is a prison network that might be a fair comparison.
[1] http://www.amazon.com/Linked-The-New-Science-Networks/dp/073...
[1] http://www.amazon.com/Linked-The-New-Science-Networks/dp/073...
Interesting visualization. I'm wondering where this data is coming from. You could probably construct a similar network out of all of the character pages on http://harrypotter.wikia.com.
That's where we got our data from! This project was done with a team of 4, and everyone entered data while trying to check facts from the Wikia. I realize that there are some edges that are missing and some other characters, but we hoped that this subset of characters and relations would be a good approximate of the real harry potter universe.
If you want to take a look at the data: https://github.com/efekarakus/potter-network/tree/master/dat...
If you want to take a look at the data: https://github.com/efekarakus/potter-network/tree/master/dat...
It'd be really particularly interesting to understand the connections themselves with some kind of context.
I'd also really like to see things like a subcollection of the network, and how they would interact if they were put into a discrete network.
Cool stuff!
I'd also really like to see things like a subcollection of the network, and how they would interact if they were put into a discrete network.
Cool stuff!
If you are referring to ego networks (the connections between your immediate connections), the visualization gets part way there. If you hover over a node, it pops up a character bio. If you click a node, it shows the immediate peer connections, but it does not go to the next step and show the connections between the people you are connected to. Analytically, that is a great social network analysis concept, since you can determine the balanced triads and determine what needs to happen to either unbalance a network (and cause the relationships to unravel) or balance a network and make it stable.
Hey thanks a lot for the feedback!
The yellow edges represents allies (for example: Hermione-Harry). The green edges are enemies.
That's an interesting idea. As a subnetwork, we allow the viewers to click on one of the four triads and it will only display those edges. However, for T3 and T1 there are so many edges that it's still not that easy to follow :)
That's an interesting idea. As a subnetwork, we allow the viewers to click on one of the four triads and it will only display those edges. However, for T3 and T1 there are so many edges that it's still not that easy to follow :)
What is the compared-to "real social network". Facebook or other online networks? Or derived from studying meatspace communities? I imagine the numbers are very different between the two.
Good question :) I was referring mainly to the small world phenomenon [1], there are also some online networks that have ~6 avg path length like Microsoft Messenger [2]. We were just curious how a fictional network would look like.
[1] http://en.wikipedia.org/wiki/Small-world_experiment [2] http://research.microsoft.com/en-us/um/people/horvitz/leskov...
[1] http://en.wikipedia.org/wiki/Small-world_experiment [2] http://research.microsoft.com/en-us/um/people/horvitz/leskov...
What is your reference for the average path length within a real social network? Six degrees of seperation?
Yup that's right!
Cool. I have a question in my mind that you might answer or comment on. As the social networks and internet usage grow steadily, should we expect a decrease in average path length of a social network? Will "six degrees of seperation" be updated to "X degrees of seperation" in the future where X<6?
G'day, I am one of the others who worked on this.
Study back in 2011 showed that in Facebook at least, the average degrees of separation was 3.74 (and closer to 2 or 3 for regional connections) ... so as we enter a more digitally connected world, Stanley Milgram's "Smallworld" study (that gave us the original 6 DoS metric) is contracting.
Can you tell us a bit about the tools that were used to create this?
The visuals are generated primarily using D3.js (great data visualization tool).
Source: I'm on the team that worked on this.
Edit: This project is open-sourced - Check it out at https://github.com/efekarakus/potter-network :)
Source: I'm on the team that worked on this.
Edit: This project is open-sourced - Check it out at https://github.com/efekarakus/potter-network :)