מושגי ליבה
Incorporating known uncertainty levels, such as missing link information, into community detection algorithms like Flow Stability enhances the accuracy and robustness of community detection in networks with incomplete data.
סטטיסטיקה
The researchers tested ∆FS on synthetic networks with a fixed size of N = 200 nodes.
They varied the connection probabilities within and between communities in the synthetic networks, with pin ranging from 0 to 0.35 and pout ranging from 0.1 to pcore.
The analysis of the Telegram network involved a dataset of N = 12,653 nodes and E = 6,925,865 edges.
The optimal partition for both FS and ∆FS on the Telegram network was found at a Markov time of t = 8.86 × 10−1.
The largest two clusters in the Telegram network, when analyzed with FS, comprised approximately 23% of the total nodes.