Many of the systems and phenomena we deal with may be represented as networks or
graph data structures made up of entities (‘nodes’) and their interactions, connections and relationships (‘links’).
As well, various transactions such as node behavior and interactions are accumulated as big data.
Entities (nodes)
(individuals, organizations,
communities, products and
documents, etc.)
Relationships (links)
(relationships between individuals,
relationships between organizations,
buyer-product relationships, relationships
between documents, etc.)
Network (Graph)
(Linked structure of relationship)
Social network analysis is a structural data analytical method and technology that quantitatively analyzes various relationships and interaction behaviors
to identify macroscopic and microscopic relationship patterns and the formation and change of relationships.
The structure of a network may be analyzed in terms of connection and disconnection, center and periphery, Cohesion and dispersion, Position and role, etc.,
and can be used to solve real-life problems in various business and research domains, including diffusion path analysis, influence evaluation, fraud/criminal
organization detection, and role prediction.
Connection
structure
Cohesive
structure
Central
structure
Role
structure
Graph Machine Learning
Graph Machine Learning
Input Graph
Graph Measures
Graph Machine Learning
Node Classification
Link Prediction
Graph Classification
Graph machine learning means using the relational characteristics of graphs for learning or learning graphs through graph neural networks.
Graph machine learning can classify nodes, predict links, and classify graphs.
Graphs can extract more accurate data patterns by analyzing
'relationships' to improve the machine learning models.
Since graph machine learning also utilizes the relationships or associations included in data for learning,
it has been proven to increase the accuracy and explainability of classification and prediction.
Researchers and business decision-makers can now secure even more accurate and reliable analytical results through a combination
of graph analysis, graph machine learning, machine learning and multivariate statistical analysis.