Analyzing social media data to identify markers of coordinated movements, using stance detection, and using clustering techniques
The present disclosure relates to methods for analyzing social media data using various techniques, including classifying at least one contagious phenomenon propagating on a network, providing stance detection based on the social media data, clustering the social media data, and/or generating and analyzing knowledge graph embeddings using the social media data.
1 . A computerized method comprising:
retrieving timestamped social media data indicating a plurality of social media events involving a plurality of source entities and a plurality of content entities;
generating, from the timestamped social media data, a plurality of temporal knowledge graphs, wherein each temporal knowledge graph corresponds to a different time period;
generating, from the plurality of temporal knowledge graphs, knowledge graph embeddings, wherein the knowledge graph embeddings comprise a plurality of cross-platform clusters of the plurality of source entities and the plurality of content entities; and
analyzing the knowledge graph embeddings to detect an influence operation in the social media data, wherein analyzing the knowledge graph embeddings comprises:
generating a digital fingerprint for at least one cluster of the plurality of cross-platform clusters, the at least one cluster being linked to the influence operation.
2 . The computerized method of claim 1 , wherein the social media data comprises social media data from at least two platforms, wherein the temporal knowledge graphs are cross-platform temporal knowledge graphs.
3 . The computerized method of claim 2 , wherein a user account for a first platform and a user account for a second platform correspond to a single node of a temporal knowledge graph.
4 . The computerized method of claim 1 , further comprising:
performing dimension reduction on the knowledge graph embeddings to yield reduced dimension data; and
performing density-based clustering on the reduced dimension data to yield the plurality of cross-platform clusters of the plurality of source entities and the plurality of content entities.
5 . The computerized method of claim 4 , further comprising characterizing the plurality of cross-platform clusters of the plurality of source entities and the plurality of content entities based on social media data associated with each cluster.
6 . The computerized method of claim 4 , further comprising generating an information flow metric for at least one content entity within a cluster of the plurality of cross-platform clusters, wherein the information flow metric indicates one or more of:
movement of the content entity within a cross-platform knowledge graph over time; or
engagement with the content entity within the cross-platform knowledge graph over time.
7 . The computerized method of claim 1 , further comprising predicting a future connection between a first entity of the knowledge graph embedding and a second entity of the knowledge graph embedding, wherein the prediction represents a likelihood of the first entity engaging with the second entity via a social media platform.
8 . The computerized method of claim 1 , further comprising:
repeatedly generating knowledge graph embeddings using different machine learning parameters;
comparing the knowledge graph embeddings generated using the different machine learning parameters; and
determining optimal machine learning parameters based on the comparing.
9 . The computerized method of claim 1 , wherein analyzing the knowledge graph embeddings further comprises comparing the digital fingerprint for the at least one cluster to other digital fingerprints of other clusters.
10 . The computerized method of claim 9 , wherein analyzing the knowledge graph embeddings further comprises determining whether there is coordinated activity between the at least one cluster and the other clusters based on the comparison.
11 . The computerized method of claim 1 , wherein generating the digital fingerprint for the at least one cluster comprises:
receiving annotation data characterizing the at least one cluster linked to the influence operation; and
clustering the annotation data.
12 . The computerized method of claim 1 , wherein generating the digital fingerprint for the at least one cluster comprises utilizing stylometry data to detect coordinated behavior or disinformation.
13 . The computerized method of claim 1 , wherein generating the digital fingerprint for the at least one cluster comprises averaging embedding coordinates across entities within the at least one cluster.
14 . The computerized method of claim 1 , wherein generating the digital fingerprint for the at least one cluster comprises utilizing a geolocation for entities within the at least one cluster.