Holistic Influence Maximization

Abstract

The steady growth of graph data from social networks has resulted in wide-spread research in finding solutions to the influence maximization problem. In this paper, we propose a holistic solution to the influence maximization (IM) problem. (1) We introduce an opinion-cum-interaction (OI) model that closely mirrors the real-world scenarios. Under the OI model, we introduce a novel problem of Maximizing the Effective Opinion (MEO) of influenced users. We prove that the MEO problem is NP-hard and cannot be approximated within a constant ratio unless P=NP. (2) We propose a heuristic algorithm OSIM to efficiently solve the MEO problem. To better explain the OSIM heuristic, we first introduce EaSyIMthe opinion-oblivious version of OSIM, a scalable algorithm capable of running within practical compute times on commodity hardware. In addition to serving as a fundamental building block for OSIM, EaSyIM is capable of addressing the scalability aspectmemory consumption and running time, of the IM problem as well.Empirically, our algorithms are capable of maintaining the deviation in the spread always within 5% of the best known methods in the literature. In addition, our experiments show that both OSIM and EaSyIM are effective, efficient, scalable and significantly enhance the ability to analyze real datasets. Categories and Subject Descriptors MOTIVATIONGrowth and pervasiveness of online social networks is no longer a new phenomenon. They have become an integral part of the dayto-day life of almost every Internet user. Their wide-spread reach * The first two authors have contributed equally to this work.Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. with its applicability in solving the above mentioned problems and beyond, has thus, been one of the most widely studied problems over the past decade. This problem is to identify a set of seed nodes so that the overall spread of information in a network, which is the potential collective impact of imparting that piece of information to these nodes, is maximized. Given that the objective of IM is to maximize the spread of information about a content, which can be a product, person, event and many more, an important aspect of this problem is the underlying information diffusion model. A diffusion model defines the dynamics of information propagation and also controls the way this information is perceived by the nodes in a network. Nevertheless, a substantially large fraction of the literature in this field has focussed on devising efficient and scalable algorithms [13, 17-19, 21-23, 28, 29, 31-33, 35, 42, 46] for IM using the classical information diffusion models [32,44]. The two fundamental diffusion models p...

Jointly Complementary&Competitive Influence Maximization with Concurrent Ally-Boosting and Rival-Preventing

“…Numerous variants of these two models are proposed by considering real world applications. To name some, time constrain models [13,16] that introduce the influence spread delay and critical time, location-aware models [20,30,31] that consider the offline location context, topic-aware models [2,9] that introduce topic models into influence spread, opinion-aware models [10,15] that assume the users possess different opinions, adaptive or multiround models [33,35,41,44] that assume the influence spread results can be observed dynamically. These extensions all consider the single agent scenario.…”

“…ISSN 2150-8097. doi:XX.XX/XXX.XX and LT model are simple and effective, with both empirical interpretability and theoretical properties, i.e., monotonicity and subomdularity. Therefore, subsequent research works focus on modifying these two models for variety of application requirements, e.g., topic-aware influence model [2], delay and critical time based influence model [13], offline location-aware model [20], multi-round model with feedback collected adaptively [35],opinion-aware model [10,15] and profit and cost aware model [1,34,36], etc.…”

Efficient presolving methods for the influence maximization problem

“…This could be much smaller than the complexity of a direct implementation of the separation of Benders optimality cuts (25), which is ( ∑ i∈ |( 𝜔 , i)|), especially when •  𝜔 's number of SCCs is much smaller than its number of nodes, or • the number of elements in the reachability sets of the nodes in  𝜔 are much smaller than that in  𝜔 .…”

Real-Time Influence Maximization in a RTB Setting

“…-Constant Each weight e ij has a constant probability. In most solutions [4,9,12,13,15,19], p is set at 0.01 or 0.1. Some define p ∈ [0.01, 0.1] [5,27].…”

“…Approximation algorithms such as EaSyIM [12], IRIE [18], SIMPATH [16], LDAG [6] or IMRANK [7], SSA-Fix [17] offer heuristics to compute (S) . Instead of computing the union of all paths as indicated in Eq.…”