Semantic Search
Retrieval that works from meaning, context and synonymy rather than from literal string matching.
Analogy Mental model
A friend asks how old that handsome guy from Titanic is now. You do not need the name to know they mean Leonardo DiCaprio. The reference is enough.
In practice What you actually do
Stop counting keyword density. Write the way a knowledgeable person explains the subject and let related terms appear because the subject requires them. Apple, fruit, iPhone and nutrition can coexist on a page about orchards, because the engine reads the relationships rather than the repetition.
Where people get burned Pitfalls
The habit of one page for one exact phrase is a leftover from before language models read whole passages. A page written for a phrase reads like it was written for a machine, and increasingly it is judged that way.
This entry was written in English from the Chinese original, not machine-translated. Read the Chinese version →