374 100,000 + Zhihu BIG V (I) : mutual attention

Two months ago, Toutiao signed more than 300 Zhihu BIG V’s, which aroused widespread concern and discussion. Specifically, we can see: How do you view the rumor that Toutiao signed more than 300 Zhihu big V’s in one breath?

Now, though the heat has died down, some confusion remains: how many Big Vs are zhihu (here “100,000 followers as a standard”), and who are they? How about the mutual attention? Put does not exist close and distant, and even close to the situation……

Just as there are people, there are rivers and lakes. Mining the above data, we may be able to see a different Zhihu BIG V ecosphere.

This paper takes Zhang Jiawei, the no.1 V of Zhihu, as the seed ID, climbs 85 users that he follows, then climbs their respective follow lists in turn, several layers down, and obtains hundreds of thousands of user IDS and mutual follow information, among which 374 people have more than 100,000 followers and 4139 people have more than 10,000 followers. The picture below is the user that Zhang pays attention to:

Continue drawing with RAWGraph as checkers:

Two months ago, I also climbed Zhihu data: “Climb Zhang Jiawei 138W + Zhihu Followers: Data Visualization”. This time, I found that anti-crawling has become much stricter, so I will not expand it here.

After obtaining the data of 374 Zhihu Big Vs and 25,090 pieces of attention, we still chose to use the open source network analysis and visualization wizard “Gephi” to draw the attention map of big Vs.

Since it has been used once before, the operation process is basically the same as last time, so it will not be described again. You can follow the English operation step by step:

Gephi mapping weibo Forwarding Map: A Case study of “@ Wife and Child in Heaven”

GEPHI — Introduction to Network Analysis and Visualization

Although this study focused from 160,000 concerns of 10,000 + small V to 25,000 concerns of 100,000 + big V, it was still a bit intensive because there were only 374 big V participants, 67 concerns per capita.

After running the algorithm, the network graph is somewhat separated, but the final result graph is still not very satisfactory as last time. Less idle talk, more warning.

Here’s a GIF to see how the Big Vs are glued together in the beginning:

After running for a long time, the graph no longer has obvious changes:

See the final result is such a mass of network is also a mouthful of old blood spit on it:

All nodes:

A small number of nodes and edges, the wheel brother @vczh has appeared, please take a stroll:

Add nodes and edges step by step:

In the central area, the situation of concern (the number of edges) gradually increases:

A glance at the above Cheng Hao: the count in the castle. R.I.P. :

Also at the bottom:

The resulting diagram looks something like this. Because attention and attention mixed together; The network is not dispersed enough, so it may be necessary to continue to screen out more elite and less data to draw a more intuitive map. This time, we can first play the simple map and make specific interpretation by ourselves. You can still see that the big Vs that like each other are actually very close together in the graph.

In addition, I drew seven colors, probably because I was once again made to vomit blood by Gephi. I could only fill in a few colors to comfort myself. Escape… Let’s not put the pit flow.

No analysis, no interpretation, no summary. Just run. Want zhihu data, want Gephi (schrodinger’s official website download), want chart……