Damping factor

As explained in Graph Algorithms for Data Science

A damping factor is the PageRank parameter representing the probability that the random-surfer process follows an outgoing relationship rather than teleporting to a random node.

The damping factor is the probability that a PageRank random surfer follows an outgoing relationship instead of teleporting to a randomly chosen node. Its complement, 1 minus the damping factor, is the teleportation probability. The typical value is 0.85, so the surfer follows links 85% of the time and jumps randomly 15% of the time. section 5.4.1

The damping factor affects PageRank in several practical ways:

• Prevents dead ends from leaking PageRank out of the network by allowing teleportation to any node, including the current node. section 5.4.1
• Adds a uniformly distributed random-jump contribution to each node during the standard PageRank calculation. section 5.4.1
• Produces a baseline score of 0.15 for nodes with no incoming relationships when the default damping factor is 0.85. section 5.6

PageRank on a follower network

A follower graph uses PageRank to identify influential users. With the default damping factor of 0.85, the algorithm combines rank passed through follower relationships with random teleportation, then writes scores to User nodes as followerPageRank. section 5.4.1 section 5.6

Figure 5.18 Simplified PageRank calculation based on network flow

Shows the simplified PageRank calculation as network flow between linked nodes and teleportation. section 5.4.1

How Graph Algorithms for Data Science covers it

How deep it goes

Explains the probability interpretation, complement, default value, uniform jumps, and protection against dead ends.

Examples

Uses a web-surfer analogy, a follower network, and the default value 0.85 producing 0.15 for nodes without incoming relationships.

What you can do after

Apply it via PageRank to interpret influence scores while accounting for random jumps and dead-end nodes.

How it connects

Connects PageRank with teleportation, dead ends, influence scoring, and Personalized PageRank.

liveBookmanning.com

Read Graph Algorithms for Data Science on liveBook Get the book at manning.com

Related concepts

• PPRconfigures · Personalized PageRank is a graph-ranking algorithm that biases random-walk restarts or teleports toward specified nodes or a personalization distribution.
• PageRankconfigures · PageRank is a link-analysis algorithm that estimates the relative importance of nodes in a directed graph from the structure and importance of incoming links.

This explainer was generated by Manning's knowledge graph from the text of Graph Algorithms for Data Science. The section links point to the parts of the book it draws on.