Text based on automatic Optical Character Recognition processes. Please use the PDF version for legal matters

CLAIMS:

1. A method predicting and ranking edges from a node from a selected set S of network nodes, comprising the steps:

finding a!! non-cyclic paths between the node from S and any other node from another seiected set T of network nodes, where edges between the nodes have assigned weights that represent the degree of relationships etween the nodes, such as, but not limited to, similarity, degree of interactions, number of shared properties and so on, where weights are in the range (0,1] with 1 representing the strongest relationship;

ce score using the e uation;

where g is a non-decreasing function of the path length; and

ranking all such edges based on the obtained confidence score, where the rank improves with the increasing confidence score.

2. A method for predicting potential new edges in a network, comprising;

defining a set S of nodes s_{(} and a set T of nodes t where ail the nodes are in the network, where none of the nodes s_{s} of set S defines an edge with any of the nodes ¾ of set T, where each given edge of the network has a respective assigned weight representing a degree of relationship between two respective nodes connected by the given edge, where weights are in the range (0,1] with 1 representing the strongest relationship;

for each node s_{(} finding ali non-cyciic network paths between node S; and each node t_{{};

nfidence score using the equation:

where g is a non-decreasing function of path iength; and

ordering the confidence scores according to value to thereby rank potential edges between nodes Ss and nodes % where the rank improves with the increasing confidence score.