Why are text embeddings useful for search?
Why C is the answer
They place similar meanings near each other as numbers
Embeddings represent meaning as vectors, so similar content ends up close together and can be found by distance.
In detail
- Embeddings place similar meanings near each other as numbers, so search can find them.
- Keyword search alone would miss this match.
Use case
An illustrative scenario, not a real incident.
A user searches for "cancel my plan", and the help article titled "End your subscription" is found, because the meanings are close.
See it in different ways
mindmap
root((Generative AI))
Token
Unit of text
Drives cost
Embedding
Meaning as numbers
Chunking
Focused passages
Hallucination
Confident and false
Temperature
Varietyflowchart LR S0["Text"] --> S1["Embedding model"] --> S2["Vector"] --> S3["Nearest matches"]
Key point: Embeddings make meaning searchable.
Why the other options are wrong
- A. They encrypt documents at rest using a model key
- They do not encrypt.
- B. They translate text between languages before search
- They do not translate.
- D. They compress files so storage costs less overall
- They are not for compression.