Early Exploration Areas

Graph-based problems are widespread and often best understood or solved using graph representations. While social graphs have dominated social network analysis over the past decade, knowledge graphs have emerged as a powerful way to model entities and relationships, driving productivity gains across teams. Embedding effective knowledge graphs into products can be a strategic differentiator, enabling advanced, market-leading capabilities.

Graph Neural Networks (GNNs) are an emerging and impactful class of neural networks designed specifically for graph-structured data, with strong potential to address complex societal problems. They enable efficient node-, edge-, and graph-level predictions and have achieved state-of-the-art performance in classification tasks. By directly operating on graphs, GNNs have also catalyzed the broader field of graph representation learning.