Proximity measures for link prediction based on temporal events

Age-series based link prediction in evolving disease networks

“…As can be seen from Fig. 7, the best AUC values were obtained at the (25)(26)(27)(28)(29)(30) interval of the case changing rate for the age-series with small size. AUC values are relatively small for the age-series with large size and the best ones were observed at the (45-55) interval of the case changing rate.…”

“…In the present work, we propose a novel link prediction method based on unsupervised strategy that considers the evolving structure of a weighted network, such as disease network, and evolving cases concerning pairs of nodes in the network. The goal of the paper is to merge Homans' idea [30] that the strength of a connection between two nodes is proportional with how often they interact with one another and Newman's method [31] that the larger the number of common neighbors between two nodes, the higher is their probability to be connected in the future [28]. Since the weighted network is used in this paper, we modified Newman's method in the form of "the larger the total weights of common neighbors between two nodes, the higher is their probability to be connected in the future".…”

“…For this purpose, time series were first built for each pair of nodes. Here, each series observation is the frequency of occurrence of links between the nodes during a specific interval [28]. All of the studies done in this area consider time concept as interval.…”

“…Brigmann et al [25] proposed a network with temporal information for discovering association rules explaining the network evolution. The other approaches proposed for link prediction handles the task as a time series forecasting problem [26][27][28]. For this purpose, time series were first built for each pair of nodes.…”

Unsupervised link prediction in evolving abnormal medical parameter networks

“…Brigmann et al proposed to mine network with temporal information for discovering association rules explaining the network evolution [17]. The other approaches proposed for link prediction handles the task as a time series forecasting problem [18][19][20]. In those studies, the time series models were used to predict proximity values.…”