Modern Graph Databases - David Ostrovsky - NDC Oslo 2026

Value of the Information & Strength of the Argument

The talk delivers substantial value by presenting cutting-edge developments in graph databases that are often overlooked. The speaker’s demonstrations in Jupyter notebooks with embedded analytical databases illustrate real-world applicability, making the arguments concrete and actionable. He effectively argues that graph databases offer unique advantages for complex relationship queries, and the contrast between traditional OLTP (Neo4j) and new embedded analytical systems is well articulated. The fraud detection example, while simplified, convincingly shows how graph queries can uncover patterns that are awkward or inefficient in SQL. The argumentation is solid, grounded in practical experience, and the progression from fundamentals to advanced use cases builds a coherent narrative.

Scientific Rigor, Source Quality, Title Accuracy

The speaker shows technical rigor by carefully explaining concepts and his experience at Meta lends credibility. However, he does not provide formal citations for most claims; for instance, he mentions DB-Engines ranking and SQLite installations without exact data. The sources cited in the description are exclusively conference and organizer pages, which do not serve as references for the technical content. The title aligns perfectly with the content, and the talk does not deviate from its promise. No audience comments were provided for analysis, so public reception trends cannot be assessed.

The title accurately reflects the content, which comprehensively surveys modern graph database technologies and their applications.

7/10

The speaker demonstrates practical expertise and covers multiple modern graph database technologies with live demos. However, some claims lack explicit citations and a few references (e.g., DB-Engines rankings) are mentioned without precise data.

Markers derived by PSI from the transcript: the creator did not define chapters.

• 00:00Introduction: speaker introduces topic and describes his 10-year history with graph databases at NDC.
• 05:00Why graph data matters: explanation of interconnected entities and limitations of relational databases.
• 10:00Storage models: property graphs vs RDF triples, and introduction of Cypher and SPARQL query languages.
• 15:00Neo4j and popularity: DB-Engines ranking and why Neo4j remains the primary OLTP graph database.
• 20:00Embedded analytical graph databases: concept and examples like LadybugDB (formerly KuzuDB).
• 25:00Live demo: querying LadybugDB with Cypher, including pattern matching and variable-length traversals.
• 30:00Fraud detection use case: demonstrating graph cycles for money laundering detection with Python and LadybugDB.
• 35:00Graph processing engines: discussion of tools like PuppyGraph and their differences from graph databases.
• 40:00AI agents and knowledge graphs: building a simple agent using FalkorDB to illustrate the role of knowledge graphs in GenAI.

• NDC Conferences — Conference organizer website mentioned in the video description.
• NDC Oslo — Specific conference page for the event where the talk was recorded.