AI 용어집
인공지능 완전 사전
HITS Centrality
Algorithm that simultaneously calculates two scores: hubs (nodes pointing to good authorities) and authorities (nodes pointed to by good hubs). Initially used for web page ranking.
Flow Centrality
Measure based on the maximum flow that can pass through a node considering all source-destination pairs. Evaluates a node's ability to facilitate resource transfer in the network.
Subgraph Centrality
Quantifies the importance of a node by counting the number of closed walks passing through that node. Uses the diagonal of the adjacency matrix exponential to capture local participation.
Load Centrality
Extension of betweenness centrality that considers all shortest paths without distinction. Measures the total number of shortest paths passing through a node between all pairs.
Harmonic Proximity Centrality
Variant of proximity centrality that uses the harmonic mean of distances rather than the sum. More suitable for disconnected graphs as it naturally handles infinite distances.
Diffusion Centrality
Measures a node's importance based on its ability to diffuse information or influences in the network. Combines paths of different lengths with specific weights.
Alpha Centrality
Generalization of Katz centrality that includes an alpha tuning parameter to control the influence of longer paths. Allows adjusting the sensitivity of the measure to distant connections.
Trade-off Centrality
Approach that combines multiple centrality measures into a single balanced metric. Aims to capture different aspects of a node's importance in a single composite value.
Percolation Centrality
Dynamic measure that evaluates how a node's centrality changes during percolation or propagation processes. Particularly useful for studying the robustness and vulnerability of networks.
Eccentricity Centrality
Defines the importance of a node by its maximum distance to the farthest node in the graph. Nodes with low eccentricity are central and geometrically well positioned.
Stress Centrality
Counts the total number of shortest paths passing through a node, similar to betweenness but without normalization. Identifies potential congestion points in the network.
Radial Centrality
Measure based on the inverse of the sum of inverse distances from a node to all others. Evaluates how close a node is to the opposite periphery of the network.
Clustering Centrality
Evaluates the importance of a node based on its participation in triangles or dense clusters. Nodes with high clustering centrality promote local network cohesion.
Community Centrality
Measures the importance of a node in the community structure of the network. Combines intra-community and inter-community factors to evaluate bridge or leader roles.