Watch: DocumentChunker
Chrome's internal Blink algorithm that recursively walks a page's DOM layout tree to group text nodes into semantically cohesive passages of up to 200 words.
Transcript
Inside Google Chrome's rendering engine, a quiet tool called DocumentChunker is changing how the browser understands web pages. Instead of treating a page as one giant, flat block of text, this algorithm digs into the document's structure to group related text together.
It works by scanning the page layout, combining neighboring text and sub-sections into meaningful, semantically cohesive passages. Each of these chunks is kept to a maximum of about two hundred words. By doing this work directly on your device where the page is actually rendered, the browser can make sense of content in real time.
DocumentChunker acts as a vital building block for other Chrome systems. It helps generate annotated page content and creates the precise, passage-level units that artificial intelligence models use later on. When an AI needs to cite a specific source on a page, it relies on these exact grounding chunks to point to the right information.