The Time-Series Link Prediction Problem with Applications in Communication Surveillance

Abstract

T he ability to predict linkages among data objects is central to many data mining tasks, such as product recommendation and social network analysis. Substantial literature has been devoted to the link prediction problem either as an implicitly embedded problem in specific applications or as a generic data mining task. This literature has mostly adopted a static graph representation where a snapshot of the network is analyzed to predict hidden or future links. However, this representation is only appropriate to investigate whether a certain link will ever occur and does not apply to many applications for which the prediction of the repeated link occurrences are of primary interest (e.g., communication network surveillance). In this paper, we introduce the time-series link prediction problem, taking into consideration temporal evolutions of link occurrences to predict link occurrence probabilities at a particular time. Using Enron e-mail data and high-energy particle physics literature coauthorship data, we have demonstrated that time-series models of single-link occurrences achieve comparable link prediction performance with commonly used static graph link prediction algorithms. Furthermore, a combination of static graph link prediction algorithms and time-series models produced significantly better predictions over static graph link prediction methods, demonstrating the great potential of integrated methods that exploit both interlink structural dependencies and intralink temporal dependencies.

Link prediction in complex networks: A survey

“…For example, it is impossible to predict whether and when two authors will collaborate again in co-authorship network. Addressing this problem, Huang and Lin [165] proposed a time-series link prediction approach considering the temporal evo-lutions of link occurrences, which is more appropriate for dealing with the link prediction problem in evolving networks, such as online social networks. Another way to involve time information is inspired by the fact that older events are less likely to be relevant to future links than recent ones.…”

Advances in Intelligent Data Analysis XIII

“…Huang and Lin [7] implemented an approach that considers the temporal evolution of link occurrences within a social network to predict link occurrence probabilities at a particular time in the future. Sun [15] proposed a meta path-based model to predict the future co-author relationships in a bibliographic network.…”

Proximity measures for link prediction based on temporal events

“…Recent activity between common neighbors may be more important than older activity. Also, static approaches are appropriate to investigate whether a certain link will ever occur in a network but they are less useful, for instance, to applications for which the prediction of repeated link occurrences are of interest (Huang & Lin, 2009).…”

“…An alternative approach for deploying temporal information is to treat link prediction as a time series forecasting problem (Huang & Lin, 2009;Potgieter et al, 2009;Qiu, He, & Yen, 2011;Soares & Prudêncio, 2012). Huang and Lin (2009) built a time series for each pair of nodes, in which each series observation is the frequency of occurrence of links between the nodes during a specific time period.…”