Overview

Adamic-Adar index is a node similarity metric defined based on the structured information of the internet, this is how it differs from Jaccard similarity (semi-structured information). AA index uses the weights of the common neighbor nodes of two nodes as the similarity of the two nodes, it was proposed by Lada A. Adamic and Eytan Adar in 2003, related materials are:

• L.A. Adamic, E. Adar, Friends and Neighbors on the Web (2003)
• Adamic-Adar Index

Basic Concept

Node Weight

In AA index, the weight of a node x is defined as the reciprocal of the logarithm based on 10 of the size of the node's neighbor set N(x):

Weight of the yellow node in the graph below is: 1/(log4) = 1.6610, weight of the green node is: 1/(log3) = 2.0959.

AA Index

AA index uses the sum of weights of the common neighbors of two nodes to determine their closeness. It is calculated by the following formula:

where N(x) and N(y) are neighbor sets of x and y respectively, u is the common neighbor of x and y. The larger the value of AA(x,y) is, the closer the two nodes are, value of 0 indicates that the two nodes are not close.

Still taking the previous graph as an example, AA index of the blue and red nodes is the sum of weights of the yellow and green nodes, which is 1/(log4) + 1/(log3) = 3.7569.

Special Case

Lonely Node, Disconnected Graph

There is no edge between lonely node and any other nodes in the graph, the algorithm does not calculate AA index between lonely node and any other node, nor does it calculate AA index between two nodes in different connected components.

Self-loop Edge

The algorithm ignores all self-loop edges when calculating neighbor nodes.

Directed Edge

For directed edges, the algorithm ignores the direction of edges but calculates them as undirected edges.

Results and Statistics

Take the graph below as an example, run the algorithm in the graph:

Algorithm results: Calculate AA index for node 3 and other nodes, return node1, node2 and num

Algorithm statistics: N/A

Command and Configuration

• Command: algo(topological_link_prediction)
• Configurations for the parameter params():

Algorithm Execution

Task Writeback

1. File Writeback

Example: Calculate AA index of node UUID = 3 and all other nodes, write the algorithm results back to file named aa

algo(topological_link_prediction).params({
  uuids: [3],
  uuids2: [1,2,4,5,6,7]
  }).write({
  file:{ 
    filename: "aa"
  }
})

2. Property Writeback

Not supported by this algorithm.

3. Statistics Writeback

This algorithm has no statistics.

Direct Return

Example: Calculate AA index of node UUID = 3 and UUID = 4, define algorithm results as alias named similarity and return the results

algo(topological_link_prediction).params({
  uuids: [3],
  uuids2: [4],
  type: "Adamic_Adar"
}) as similarity 
return similarity 

Streaming Return

Example: Calculate AA index of node UUID = 1 and UUID = 5,6,7, return the results in the descending closeness score

algo(topological_link_prediction).params({
  uuids: [1],
  uuids2: [5,6,7],
  type: "Adamic_Adar"
}).stream() as com 
return com order by com.num desc 

Real-time Statistics

This algorithm has no statistics.