Link segmentation entropy for measuring the network complexity

Measuring the network complexity has been addressed in many studies in graph theory. In the context of complex networks, a class of complexity measures has been proposed based on structure entropy. Although the proposed measures can quantify the complexity of different networks, they are mainly focused on the nodes structural properties and ignore the links information. This is while the link analysis plays a crucial role in studying and comprehension of different types of complex networks. In this paper, we propose to employ the similarity-based link prediction measures to capture the links information and quantify it as an entropy measure. The findings of the experimental study based on several synthetic as well as real-world networks demonstrate the validity and effectiveness of the proposed complexity measure.