Link prediction in complex networks based on an information allocation index

- A generalized information-theoretic model has been proposed.
- The concept of information allocation has been proposed and introduced to link prediction.
- A new index with better overall performance has been designed.

An important issue in link prediction of complex networks is to make full use of different kinds of available information simultaneously. To tackle this issue, recently, an information-theoretic model has been proposed and a novel Neighbor Set Information Index (NSI) has been designed. Motivated by this work, we proposed a more general information-theoretic model by further distinguishing the contributions from different variables of the available features. Then, by introducing the resource allocation process into the model, we designed a new index based on neighbor sets with a virtual information allocation process: Neighbor Set Information Allocation Index(NSIA). Experimental studies on real world networks from disparate fields indicate that NSIA performs well compared with NSI as well as other typical proximity indices.