[Paper Review] A Survey on Location-Driven Influence Maximization

Taotao Cai, Quan Z. Sheng+6 more|arXiv (Cornell University)|Apr 17, 2022

Human Mobility and Location-Based Analysis4 citations

TL;DR

This survey provides a comprehensive analysis of location-driven influence maximization (IM), categorizing it into Online Location-Aware IM (OLIM) and Offline Billboards IM (OBIM). It reviews diffusion models, solution frameworks, and acceleration techniques, offering a fine-grained taxonomy and identifying future research directions in dynamic networks, non-submodular models, and seed selection under real-world constraints.

ABSTRACT

Influence Maximization (IM), which aims to select a set of users from a social network to maximize the expected number of influenced users, is an evergreen hot research topic. Its research outcomes significantly impact real-world applications such as business marketing. The booming location-based network platforms of the last decade appeal to the researchers embedding the location information into traditional IM research. In this survey, we provide a comprehensive review of the existing location-driven IM studies from the perspective of the following key aspects: (1) a review of the application scenarios of these works, (2) the diffusion models to evaluate the influence propagation, and (3) a comprehensive study of the approaches to deal with the location-driven IM problems together with a particular focus on the accelerating techniques. In the end, we draw prospects into the research directions in future IM research.

Motivation & Objective

• To systematically classify and analyze existing research on location-driven influence maximization based on application scenarios.
• To review and compare influence diffusion models used in OLIM and OBIM problems.
• To summarize framework algorithms and acceleration techniques for solving OLIM and OBIM problems efficiently.
• To identify critical challenges and outline promising future research directions in dynamic, non-submodular, and budget-constrained IM settings.

Proposed method

• Categorizes location-driven IM into two main lines: OLIM (online social networks with user locations) and OBIM (offline billboards for influence spread).
• Reviews key influence diffusion models such as the Independent Cascade (IC) and Linear Threshold (LT) models adapted for spatial and temporal dynamics.
• Analyzes framework algorithms including greedy heuristics, randomized local search, and sketch-based methods like Reverse Reachable (RR) sets for scalability.
• Evaluates acceleration techniques such as pruning, sampling, and index-based maintenance to improve runtime efficiency.
• Proposes a taxonomy based on problem type, diffusion model, and algorithmic approach for systematic comparison.
• Introduces the concept of temporal link prediction and dynamic sketch indexing to address evolving network structures.

Experimental results

Research questions

• RQ1 How can influence maximization be effectively extended to incorporate spatial and temporal user location data?
• RQ2 What are the key differences and similarities between OLIM and OBIM in terms of problem formulation and solution techniques?
• RQ3 How do existing diffusion models capture location-aware influence propagation in social and physical spaces?
• RQ4 What acceleration techniques are most effective in scaling influence maximization algorithms for large-scale location-driven networks?
• RQ5 What are the major open challenges in handling dynamic network evolution, non-submodular influence functions, and user reluctance in real-world IM applications?

Key findings

• The OLIM and OBIM frameworks are well-established, with OLIM focusing on seed user selection in dynamic social networks and OBIM on optimal billboard placement for out-of-home advertising.
• Diffusion models such as IC and LT are adapted for location-aware influence spread, but their effectiveness depends on accurate spatial-temporal modeling.
• Greedy and randomized local search methods are widely used, but they face limitations in non-submodular settings, where convergence to global optima is not guaranteed.
• Acceleration techniques like reverse reachable sketching and pruning significantly reduce computational cost, enabling scalability to large networks.
• Temporal link prediction and dynamic sketch indexing are promising approaches to address network evolution, though still underexplored in current literature.
• Future research should focus on dynamic networks, non-submodular models, and modeling user reluctance through edge-based influence rather than node-based seeding.