Link Prediction in Co-Authorship Network under Fuzziness and Application in Biomedical Analysis

We aim to predict links in fuzzy social networks, where the existing methods based on common neighbors of two nodes are not effective. These methods are local measures that only work when the shortest distance between two nodes is less than or equal to two. Our method can handle cases where the shortest distance is between three and five. We define the concepts of link strength and path strength in a network and propose an algorithm for predicting links. We illustrate our method with a numerical example in a co-authorship network and discuss application areas in biomedical.