Semantic Evolution

Semantic evolution studies how word meanings and concepts change over time, aiming to understand the underlying mechanisms and patterns of this change across various domains. Current research focuses on applying techniques like graph autoencoders and distributional semantic embeddings to model semantic shifts in diverse contexts, including social media, language models, and scientific literature. These analyses reveal insights into phenomena such as subdiffusive semantic drift in languages and the layered representation of lexical semantics in large language models, with implications for applications ranging from rumor detection to few-shot learning and long-term autonomous navigation.