Many columns in this series have focused on research advice and have provided excellent tools for thinking about the transition from junior to senior researcher. So let me instead talk about the data management research system that we find ourselves a part of, whether academic, industrial, or governmental. Let me call it Big Research . Congratulations! You have mastered Big Research and Big Research has changed you. You have figured out how to navigate this complex and challenging discipline and are now one of us! Now is the time to pause and remember what you hated about the process of getting here and reflect on whether all those hurdles were good hurdles. Many - no, most - made you a better researcher, scientist, communicator, and mentor, but did they all? Don't be someone who says, ''I had to run this unproductive gauntlet, so therefore everyone else should.''
The International Workshop on Transforming Graph Data (TGD) ran for two editions, co-located with EDBT/ICDT in Barcelona in March 2025 and in Tampere in March 2026. It was motivated by an asymmetry in the graph data management ecosystem. Storage, indexing, and querying have matured considerably, while the mechanisms needed to evolve, integrate and reason over graph data remain comparatively ad hoc. The program comprised 12 papers, four keynotes and one panel, spanning declarative and logic-based transformation languages, schema evolution and discovery, interoperability between property graphs and RDF, hybrid graph and time-series models, and transformation-aware summarization and lineage. This report summarizes the contributions, the recurring themes, and the agenda that emerged.
Motivation : Modern data analytical workloads – from graph processing to machine learning – have evolved significantly, increasingly combining JOIN with additional relational-algebra operators and aggregate functions (e.g., PROJECT and COUNT). Traditionally, database query processing has been dominated by discrete, logicbased primitives such as for-loops [9, 10, 13], tree decompositions [4], and data partitioning [6, 8], which are fundamentally combinatorial . However, the landscape changes when examining graph analytics, where algebraic algorithms – which treat query processing as a series of high-dimensional tensor or matrix operations – have achieved strictly lower theoretical complexity than the best-known combinatorial counterparts. Fast matrix multiplication (FMM) bridges this gap. As modern data systems increasingly integrate dedicated linear algebra kernels (e.g., for ML), repurposing these optimized algebraic engines for classical query processing offers immense potential. This drives the core insight of our research: a comprehensive exploration of FMM within database query engines, bringing the theoretical power of algebraic techniques to practical data systems.
Welcome to ACM SIGMOD Record's series of interviews with distinguished members of the database community. Today I'm joined by Anthony K.H. Tung, Professor of Computer Science at the National University of Singapore, where he is also the AI lead at the NUS Artificial Intelligence Institute and Director of the NUS-Google Joint Research and Innovation Center. Among other honors, his work with the Apache SINGA team has been recognized with the 2024 ACM SIGMOD Systems Award. Anthony, welcome!
This report summarizes discussions that took place in the context of a Dagstuhl seminar1 to explore the emerging paradigm of AI-native databases designed to support agentic workflows and Retrieval-Augmented Generation (RAG) by integrating Large Language Models (LLMs) with robust database abstractions. The central synthesis of the Dagstuhl discussions is the necessary evolution of databases from passive vector stores into active data-native agentic harnesses that orchestrate AI reasoning and task-centric execution. A recurring lesson across keynotes and working groups is that achieving mission-critical reliability requires integrating AI primitives, such as embeddings, model inference, and semantic operators, as first-class citizens while maintaining traditional database guarantees like transactional consistency, explainability, and cost-aware optimization. This transition necessitates a structured approach to agent memory (spanning episodic, semantic, and procedural scales) and a shift toward adaptive, iterative execution models that go beyond static query planning to accommodate the stochastic nature of foundation models. There is a community consensus that reliability, responsibility, and security (specifically addressing compositional leakage) must be treated as end-to-end concerns supported by holistic benchmarks that evaluate the entire quality chain from retrieval to autonomous action.
