Professional Documents
Culture Documents
Social Network Analysis (SNA) Is A Quantitative Method To Examine and Measure
Social Network Analysis (SNA) Is A Quantitative Method To Examine and Measure
Y1 Y2 Y3 Y4
Reduced risk: can be used to identify and mitigate risks associated with social
networks. can be used to detect social bots and spammers.
Crime prevention: identify criminal networks and to predict where crimes are likely
to occur. This information can be used by law enforcement agencies to prevent
crimes and to apprehend criminals.
Role of GNN
To label the nodes and edges in the graph, we can use a machine learning model
that can learn the node and edge features and predict the classes. Graph neural
network (GNN) can be used to encode the node and edge features into node and
edge representations, and then use an edge classifier to predict the edge labels
based on the representations, similarly we can predict the node labels also.
To use a GNN to make predictions on social network, you we first need to represent
the network as a graph. Each node in the graph would represent a social media
account, and each edge in the graph would represent a relationship between two
accounts. The edge attributes could be used to represent the strength and type of
relationship between the two accounts.
Once the representation is made your social network as a graph, you can train a
GNN on the graph. The GNN will learn to represent each node in the graph as a
vector of numbers. This vector will capture the information about the node's
relationships to other nodes in the graph.
Once the GNN is trained, it can be used to make predictions about the nodes and in
the graph.
Problem Identification and Objectives
Identification
objectives
To show how strong the friendship is between two nodes in the network, we
classified edges into strong friendship, weak friendship, uncertain friendship,
liberal friendship.
labels the nodes to represent their roles in the network, such as leader,
follower, moderator, etc.
System methodology
System Design
The data collection and preprocessing module will collect data from social network
platforms, such as Facebook, Twitter, and LinkedIn. The data will be preprocessed
to remove noise and outliers. This may include removing duplicate data, correcting
errors, and filtering out irrelevant data.
Graph Conversion
The graph conversion module will convert the preprocessed data into a graph
format. Represent four degrees of membership: true (t), indeterminate (i), false (f),
and liberal (l). The liberal value of an edge can be used to represent the degree to
which one node trusts the other node. The membership value of an edge can be
used to represent the strength of the friendship between two nodes.
GNN Training
The GNN training module will train a GNN model on the graph. GNNs are a type of
neural network that can learn from graph data. The GNN model will be trained to
predict the trust, friendship, and role of each node in the social network.
Prediction
The prediction module will use the trained GNN model to make predictions about the
social network. For example, the prediction module can be used to predict the
following:
Testing
The system will be tested using held-out test and cross-validation to measure
accuracy. Held-out test involves splitting the data into two sets: a training set and a
test set. The model is trained on the training set and then evaluated on the test set.
Cross-validation involves splitting the data into multiple folds and training the model
on each fold. The model is then evaluated on the average performance of all the
folds.
Overview of Technologies
NumPy
Pandas
Matplotlib
Social network analysis is a field that studies the structure and dynamics of social
relationships and interactions among individuals or groups. Graphs are a natural way to
model social networks, where nodes represent entities and edges represent relations.
However, traditional graphs cannot capture the uncertainty and refusal that exist in real life
situations, such as friendship status. Therefore, we propose to use a type of graph that can
represent four degrees of membership: t (true), i (indeterminate), f (false), and l (liberal). It
can be used to model a variety of social network phenomena, such as
• Trust: The liberal value of an edge can be used to represent the degree to which one node
trusts the other node.
• Friendship: The membership value of an edge can be used to represent the strength of the
friendship between two nodes.
For example, a person may be friends with some people, uncertain about being friends with
others, not friends with some people, or liberal about being friends with anyone. These
graphs can model these different degrees of friendship by assigning different values to the
edges in the graph and GNN can then learn to predict the membership values of the edges in
the graph and classify them into strong friendship, weak friendship, uncertain friendship,
liberal friendship.
Further the nodes can also be labelled to indicate their role such as leader, follower,
moderator. GNNs are a powerful tool because they can learn the features of each node
based on the features of its neighbours and the edges between them. This allows GNNs to
learn complex patterns in the social network and to make more accurate predictions