This session will examine how graph structures can support machine-assisted puzzle-solving in the New York Times Connections game. The talk will show how puzzle elements, word relationships, and category patterns can be represented as a graph in Neo4j, giving AI systems a way to operate on links rather than isolated features.
The session will outline how to build a Neo4j-backed workflow for clustering, semantic grouping, and NLP-driven similarity scoring. It will cover how to store embeddings, word associations, and candidate categories as nodes and relationships, and how to use Cypher queries or graph algorithms to guide solver decisions.
Attendees will see how graph patterns help estimate puzzle complexity, highlight ambiguous groupings, and explain solver choices. The session will also discuss how graph insights can support game design analysis and help teams study how players interpret categories and word relationships.
I'm currently a Senior Android Games Engineer that works on the New York Times Games Team, working on the integration of games such as Wordle, Connections and the NYT Crossword, into the NYT Games Android App.
I am also an A.I. enthusiast and practitioner with multiple years experience in the field and published research that has been deployed across real world systems with both civilian to government applications.
Linked In: https://www.linkedin.com/in/shafik-quoraishee/
Free online courses and certifications. Join the 100K+ Neo4j experts.
Manuals for Neo4j products, Cypher, and drivers
Live and on-demand events, training, webinars, and demos
