Influential nodes selection: a data reconstruction perspective
Lists
Only you see this
Abstract
Influence maximization is the problem of finding a set of seed nodes in social network for maximizing the spread of influence. Traditionally, researchers view influence propagation as a stochastic process and formulate the influence maximization problem as a discrete optimization problem. Thus, most previous works focus on finding efficient and effective heuristic algorithms within the greedy framework. In this paper, we view the influence maximization problem from the perspective of data reconstruction and propose a novel framework named \textsl{Data Reconstruction for Influence Maximization}(DRIM). In our framework, we first construct an influence matrix, each row of which is the influence of a node to other nodes. Then, we select $k$ most informative rows to reconstruct the matrix and the corresponding nodes are the seed nodes which could maximize the influence spread. Finally, we evaluate our framework on two real-world data sets, and the results show that DRIM is at least as effective as the traditional greedy algorithm.
BibTeX
@inproceedings{wang2014influential,
author = {Zhefeng Wang and Hao Wang and Qi Liu and Enhong Chen},
title = {{Influential nodes selection: a data reconstruction perspective}},
booktitle = {International ACM SIGIR Conference on Research and Development in Information Retrieval},
year = {2014},
doi = {10.1145/2600428.2609464},
}Loading related content...