Machine learning has the potential to revolutionize technical writing, with embeddings emerging as the technology with the most significant impact. Embeddings enable connections to be explored at unprecedented scales, allowing comparisons between texts of any size mathematically. This concept operates in multi-dimensional spaces, with numbers in arrays representing semantic relationships. The intriguing ability of embeddings to convey semantic relationships is exemplified by the “king-man+woman=queen” equation, illustrating how these models can represent relationships intuitively. By leveraging embeddings in technical writing, novel opportunities for advancing documentation may arise, encouraging exploration of this innovative tool.
https://technicalwriting.dev/data/embeddings.html