Abstract: PageRank remains one of the most influential algorithms in network science, powering applications from web search to social network analysis. Traditionally, PageRank computations have relied heavily on random walk–based methods, which, while effective, can be computationally expensive and difficult to scale in dynamic environments. As networks grow larger and evolve rapidly, the need for parallel and dynamic algorithms becomes increasingly critical.
In this talk, Dr. Bapi Chatterjee and Dr. Piotr Sankowski from IDEAS NCBR will explore whether random walks are truly necessary for PageRank and present alternative approaches that leverage parallelism and dynamic updates. The session will highlight recent advances in algorithmic design that enable faster, more efficient computation of PageRank scores, even in networks that change continuously.
Key themes will include:
Limitations of random walk–based PageRank methods in large-scale, evolving networks.
Parallel algorithms that accelerate PageRank computation across distributed systems.
Dynamic approaches for maintaining PageRank scores as networks update in real time.
Practical applications in search engines, recommendation systems, and graph analytics.
The presentation will combine theoretical insights with practical demonstrations, showing how modern algorithmic innovations can reshape our understanding of PageRank. By questioning the necessity of random walks, the speakers will open new directions for scalable, adaptive, and efficient network analysis in the era of massive, dynamic data.