Sentence examples for multiplex links in from inspiring English sources

Our first goal is to gain insight into the geo-social structural and interaction properties of multiplex links in the multilayer social network and how they differ from other link types.

In Figure 5(C) and (D) we can observe that we are able to achieve greater predictive power using Twitter features in predicting multiplex links than Foursquare links in Figure 5(A) and in using Foursquare features in Figure 5(B), with the highest AUC scores of 0.82 and 0.84 for each set respectively.

Finally, we can see that using multilayer and geo-social features which employ both spatial and social interactions from the two heterogeneous platforms can outperform both single layer sets in predicting multiplex links (highest (mathrm{AUC} = 0.88) for Chicago).

Nevertheless, we have observed some evidence of media multiplexity manifested in the greater intensity and structural overlap of multiplex links and have gained insight into how we can utilise these properties for link prediction.

We study the three types of links as described in our multilayer model above: multiplex links across both Twitter and Foursquare, which we denote as tf for simplicity; single-layer links on Foursquare only (denoted as fo); single-layer links on Twitter only (denoted as to), and compare these to unconnected pairs of users (denoted as na).

In this work, we study the geo-social properties of multiplex links, spanning more than one social network and apply their structural and interaction features to the problem of link prediction across social networking services.

In agreement with previous studies of tie strength [20], we observe that multiplex links share greater structural similarity than other link types across network configurations and this will be s useful property in our link prediction problem.

While mentions are more discriminative between multiplex links (tf) and single-layer connectivity (to and fo), hashtags are better at distinguishing between links and non-links (na) in terms of median values.

This confirms our assumption that multiplex links are easier to identify than single layer links by using the same algorithmic set-up and shows that the strength of multiplex ties exhibited in the first part of our analysis can be used to predict links across networks.

In the present work, we set the exponents (a=2, b=1) after optimising for the exponents that maximise the difference between the median values of multiplex links (tf) and no link (na).

This indicates that both Foursquare spatial features are better at distinguishing multiplex links and native Foursquare links than other link types based on the distributions observed.