Pagerank Algorithms and Multi-Particle Environments Using Pettingzoo: Network Simulation and Validation

Abstract

This study introduces and investigates two novel algorithms, namely Topic-Specific Crawling PageRank and Motif-Based PageRank, along with their corresponding application scenarios. The former algorithm focuses on targeted web crawling, refining the collection of web pages relevant to specific topics. The latter algorithm addresses higher-order relationships in citation networks, enhancing insights into researcher influence. These algorithms are evaluated within a Multi-Particle Environment, showcasing their capability to simulate necessary scenarios and data for PageRank algorithms. However, the environment's realism is limited due to its use of virtual data. The environment offers controllability through the PettingZoo resource library, providing an efficient experimentation platform. While the Multi-Particle Environment displays potential and advantages, its deviation from real-world data should be carefully considered. The experimental methodology involves constructing a network of agents and their message interactions within a simulated environment. The resulting network undergoes the PageRank algorithm, and the resulting ranking aligns with the expected outcome, confirming the Multi-Particle Environment's compatibility with the algorithm. These experiments furnish empirical evidence for the Multi-Particle Environment's adaptability to algorithmic simulations and real-world scenarios. This research contributes to a deeper comprehension of algorithm behavior in diverse and dynamic settings, paving the way for further advancements in the field