Link prediction aims to uncover missing links or predict the emergence offuture relationships according to the current networks structure. Plenty ofalgorithms have been developed for link prediction in unweighted networks, withonly a very few of them having been extended to weighted networks. Thus far,how to predict weights of links is important but rarely studied. In thisLetter, we present a reliable-route-based method to extend unweighted localsimilarity indices to weighted indices and propose a method to predict both thelink existence and link weights accordingly. Experiments on different realnetworks suggest that the weighted resource allocation index has the bestperformance to predict the existence of links, while the reliable-route-basedweighted resource allocation index performs noticeably better on weightprediction. Further analysis shows a strong correlation for both linkprediction and weight prediction: the larger the clustering coefficient, thehigher the prediction accuracy.
展开▼
