System and method for social network analysis
A system and method for detecting interactive network of automated accounts, the interactive network of automated accounts comprising a plurality of automated accounts posting to a social media channel, the system comprising: an ingestion engine operated by a computational device for connecting to the social media channel and receiving a plurality of social media postings from a plurality of posting entities; a bot model operated by a computational device for determining whether at least one posting entity is a suspected bot; and a computer network for communication between said computational devices.
1. A system for detecting an interactive network of a plurality of automated accounts posting to a social media channel, wherein each automated account is operated by at least one social media account computational device, the system comprising: an ingestion computational device comprising a first memory for storing a plurality of first instructions and a first processor for executing said first instructions, wherein said first instructions include instructions for:
connecting to the social media channel; and
receiving a plurality of social media postings from a plurality of social media accounts;
the system further comprising an analysis computational device comprising a second memory for storing a plurality of second instructions and a second processor for executing said second instructions, wherein said second instructions include instructions for:
analyzing said social media postings according to interactions between said social media accounts; and
determining which of said posting entities are in a community according to said analyzing;
and a computer network for communication between said computational devices;
wherein said ingestion engine obtains social media messages for a plurality of social media accounts, said ingestion computational device transmitting said social media messages to said analysis computational device; and
wherein said analyzing said interactions is performed according to an algorithm selected from the group consisting of: page rank variations, Louvain community detection, K-core Decomposition for virality measurement, Triangle Count, Heuristic rules, Story-level features, Author features and Graph-level features.
2. The system of claim 1 , wherein said interactions between said social media accounts are selected from the group consisting of likes, re-posting, quote posting, commenting, replies and mentions.
3. The system of claim 1 , wherein said instructions on said analysis computational device further comprise instructions for placing said social media accounts as nodes on a graph, with said interactions between said social media accounts being represented as edges; and analyzing said interactions according to one or more features of said graph.
4. The system of claim 3 , wherein each set of interactions between each pair of nodes is decomposed to a plurality of edges, wherein each edge represents a type of interaction.
5. The system of claim 4 , wherein said edge is weighted according to a number of each type of interaction.
6. The system of claim 3 , wherein each set of interactions between each pair of nodes is decomposed to a plurality of edges, wherein each edge represents a single interaction.
7. The system of claim 3 , wherein said instructions on said analysis computational device further comprise instructions for analyzing said interactions between said nodes to determine if a node belongs to a community according to one or more graph parameters selected from the group consisting of: Minimum PageRank score of interacting nodes; Maximum PageRank score of interacting nodes; Standard Deviation PageRank score of interacting nodes; Standard Deviation Triangle Count score of interacting nodes; Stdev Triangle Count score of interacting nodes; 90th percentile of Triangle Count score of interacting nodes; 90th percentile of PageRank score of interacting nodes; 99.9th percentile of Triangle Count score of interacting nodes; or 99.9th percentile of PageRank score of interacting nodes; or a combination thereof.
8. The system of claim 1 , wherein said interactions are analyzed for automatic behavior by a social media account, such that said social media account is at least partially determined to belong to said community according to said automatic behavior.
9. The system of claim 1 , wherein said social media computational device comprises one or more of a laptop, a PC, a mobile communication device, a mobile telephone, a tablet computational device and a plurality of cloud based computational services.
10. The system of claim 9 , wherein said social media computational device comprises a support social media application for posting social media messages to said social media channel in an automated manner, wherein said analysis computational device determines an automation behavior score for a social media account associated with said social media computational device at least partially according to a characterization of said support social media application, such that said automation behavior score at least partially determines whether said social media account belongs to said community.
11. The system of claim 1 , wherein said memory of said analysis computational device comprises a first set of machine codes selected from the native instruction set for receiving instructions from said ingestion computational device, and a second set of machine codes selected from the native instruction set for executing functions of analysis engine; wherein said analysis computational device comprises an operating system and wherein said operating system determines said native instruction set; and wherein each set of machine codes is stored on a memory associated with said computational device.
12. The system of claim 1 , wherein said analyzing said interactions further comprises recursively analyzing a plurality of interacting social media accounts, to detect one or more suspect accounts that are suspected of belonging to the community; calculating a score for each suspect account; and labelling a suspect account as member of the community if said score is above a cut-off threshold.
13. The system of claim 12 , wherein said score comprises an automation score and said community comprises a plurality of automated social media accounts.
14. The system of claim 13 , wherein said automated social media accounts engage in a concerted effort to change sentiment about another social media account that is not a member of the community.
15. The system of claim 13 wherein said analyzing said interactions further comprises creating a map of interactions between members of the community.
16. The system of claim 15 , wherein said map of interactions is determined as a graph, wherein said graph comprises a plurality of nodes, wherein said nodes represent members of the community and links between said nodes represent interactions between said members.
17. The system of claim 1 , wherein said analyzing said interactions further comprises recursively analyzing a plurality of interacting social media accounts, to detect one or more suspect accounts that are suspected of belonging to the community; wherein members of the community are determined to interact with each other bidirectionally and with social media accounts outside of the community unidirectionally, such that said members of the community respond to social media accounts outside of the community but said social media accounts outside of the community fail to respond to said members of the community.
18. The system of claim 1 , wherein said analyzing said interactions further comprises analyzing interactions of social media accounts with each other to classify said social media accounts as influencers, repeaters, authentic users, and spammers.
19. A system for detecting an interactive network of a plurality of automated accounts posting to a social media channel, wherein each automated account is operated by at least one social media account computational device, the system comprising: an ingestion computational device comprising a first memory for storing a plurality of first instructions and a first processor for executing said first instructions, wherein said first instructions include instructions for:
connecting to the social media channel; and
receiving a plurality of social media postings from a plurality of social media accounts;
the system further comprising an analysis computational device comprising a second memory for storing a plurality of second instructions and a second processor for executing said second instructions, wherein said second instructions include instructions for:
analyzing said social media postings according to interactions between said social media accounts;
wherein said analyzing said interactions is performed according to an algorithm selected from the group consisting of: page rank variations, Louvain community detection, K-core Decomposition for virality measurement, Triangle Count, Heuristic rules, Story-level features, Author features and Graph-level features; and
determining which of said posting entities are in a community according to said analyzing;
and a computer network for communication between said computational devices; wherein said ingestion engine obtains social media messages for a plurality of social media accounts, said ingestion computational device transmitting said social media messages to said analysis computational device; and wherein said analyzing said interactions further comprises recursively analyzing a plurality of interacting social media accounts, to detect one or more suspect accounts that are suspected of belonging to the community; wherein members of the community are determined to interact with each other bidirectionally and with social media accounts outside of the community unidirectionally, such that said members of the community respond to social media accounts outside of the community but said social media accounts outside of the community fail to respond to said members of the community.
20. The system of claim 19 , wherein said automated social media accounts engage in a concerted effort to change sentiment about another social media account that is not a member of the community.