A 1.5 Million Word Semantic Network of English: The Linguistics Behind a Word Game
2025-06-03
Building a word game led researchers to construct a semantic network encompassing 1.5 million English terms. By combining human-curated thesauri, book cataloging systems, and carefully crafted LLM queries, they created a network where 76% of random word pairs connect in 7 or fewer hops. Overcoming challenges posed by superconnector words and balancing multiple ranking signals, the resulting network reveals the surprisingly close connections between English words and provides ideal parameters for game design. This research demonstrates how diverse data sources and techniques can be combined to build a semantic network that's both scientifically insightful and entertaining.
Read more
Development
semantic network