added Python scripts, did a lot of data cleaning, changed node size to be proportional to number of channels

This commit is contained in:
Tristan Lee
2022-12-15 19:42:55 -06:00
parent 88b43dfd31
commit 9b9e71fab8
9 changed files with 2915 additions and 14 deletions

View File

@@ -59,7 +59,7 @@ const ClustersPanel: FC<{
}
>
<p>
<i className="text-muted">Click a cluster to show/hide related pages from the network.</i>
<i className="text-muted">Click a cluster to show/hide related entities from the network. Cluster names were chosen to best describe most entities within the cluster, but do not always perfectly describe all entities within the cluster.</i>
</p>
<p className="buttons">
<button className="btn" onClick={() => setClusters(mapValues(keyBy(clusters, "key"), () => true))}>

View File

@@ -17,10 +17,13 @@ const DescriptionPanel: FC = () => {
This visualisation represents a <i>network</i> of{" "}
<a target="_blank" rel="noreferrer" href="https://spacy.io/usage/linguistic-features#named-entities">
named entities
</a> in English-language posts archived in a database of Telegram channels that have posted about GESARA. Each{" "}
</a> in English-language posts archived in a database of Telegram channels that have posted about the GESARA conspiracy theory. Each{" "}
<i>node</i> represents an entity, <i>edges</i> between nodes indicate that one or more posts contain both entities
.
</p>
<p>
This kind of visualization shows the ecosystem of the people, organizations, and ideas these conspiracy Telegram channels talk about, as well as the connections between them.
</p>
<p>
Some social media channels were identified by researchers from{" "}
<a target="_blank" rel="noreferrer" href="https://www.bellingcat.com/">
@@ -50,21 +53,20 @@ const DescriptionPanel: FC = () => {
sigma.js
</a>
. You can read the source code{" "}
<a target="_blank" rel="noreferrer" href="https://github.com/jacomyal/sigma.js/tree/main/demo">
<a target="_blank" rel="noreferrer" href="https://github.com/bellingcat/gesara-entity-viz">
on GitHub
</a>
.
</p>
<p>
Node sizes are related to the number of times the entity was mentioned in the database.
</p>
<p>
The network was visualized using{" "}
<a target="_blank" rel="noreferrer" href="https://gephi.org/">
Gephi
</a>. Node sizes are related to the number of channels the entity was posted about in the database.
Nodes are colored based a{" "}
<a target="_blank" rel="noreferrer" href="https://arxiv.org/abs/0803.0476">
community detection algorithm
</a>.
</p>
<p>
For visualisation purposes, edges were pruned using the{" "}
<a target="_blank" rel="noreferrer" href="https://github.com/naviddianati/GraphPruning">
Marginal Likelihood Filter