1: \begin{abstract}

2: Network (or graph) sparsification compresses a graph by removing inessential edges.

3: By reducing the data volume, it accelerates or even facilitates many downstream analyses.

4: Still, the accuracy of many sparsification methods, with filtering-based edge sampling being the most typical one, heavily relies on an appropriate definition of edge importance.

5: Instead, we propose a different perspective with a generalized local-property-based

6: sampling method, which preserves (scaled) local \emph{node} characteristics.

7: Apart from degrees, these local node characteristics we use are the expected (scaled) number of wedges and triangles a node belongs to.

8: % The scaled local characteristics involve degrees and 3-node (so far considered) subgraphs.

9: % In the context of sparsification, they are interpreted as the expected values of degrees, triangles, and wedges associated with nodes.

10: Through such a preservation, main complex structural properties are preserved implicitly.

11: We adapt a game-theoretic framework from uncertain graph sampling by including

12: a threshold for faster convergence (at least $4$ times faster empirically) to approximate solutions.

13: Extensive experimental studies on functional climate networks show the effectiveness of this method in preserving macroscopic to mesoscopic and microscopic network structural properties.

14: % Our results also highlight the potential of this method for sparsification on more generic network structures.

15: \end{abstract}

16: