From plain text to structured entity data
Search engines and AI systems increasingly rely on entities (named people, organizations, products, and
places) rather than just keywords to understand what a page is about. This tool scans your text for
capitalized name-like phrases, lets you confirm which ones are real entities and classify their type, then
generates a mentions JSON-LD block you can add to your page to make those entities explicit to
search engines.
Why this uses a pattern-based scan rather than genuine language understanding
The extractor looks specifically for capitalized words appearing where a proper noun typically would (mid- sentence, not at a sentence's start unless it continues as a multi-word capitalized phrase) - a lightweight, fast approach that catches the large majority of real names, organizations, and products, but without the overhead of a full natural-language entity-recognition model. That's why you review and confirm each candidate rather than having every flagged phrase added automatically.
Why you choose each entity's type manually instead of the tool guessing it
Telling "Apple" the company apart from "Apple" the fruit, or a person's name from a place name, generally requires real-world context a pattern-matching scan doesn't have - letting you assign the correct schema.org type (Organization, Person, Product, Place, or Thing) to each confirmed entity keeps the generated markup accurate, rather than guessing and risking mislabeled structured data.
Why this generates a mentions block rather than a full custom schema type
A mentions property on an Article schema is a broadly applicable, low-risk way to
tell search engines which named entities a piece of content discusses, without requiring you to restructure
your whole page's markup around a more specific schema type - it adds entity context on top of whatever
schema your page already uses.