Source documents are scanned images. What do you need?
Why B is the answer
OCR
Images need optical character recognition. HTML tools cannot read them.
In detail
- Scanned pages are images, so they hold no text to extract.
- Optical character recognition reads the text from the image.
- Check OCR quality on a sample, because poor scans give poor text.
Use case
An illustrative scenario, not a real incident.
A records office has decades of scanned forms. They are images, so there is no text to parse. OCR reads the text, which can then be chunked and indexed.
See it in different ways
flowchart LR A["Source documents"] --> B["Extract text"] B --> C["Filter extraneous content"] C --> D["Chunk"] D --> E["Delta table in Unity Catalog"] E --> F["Vector Search index"]
stateDiagram-v2 [*] --> Raw Raw --> Text: parse or OCR Text --> Clean: remove extraneous content Clean --> Chunked: split Chunked --> Governed: Delta table in Unity Catalog Governed --> Indexed: Vector Search Indexed --> [*]
flowchart LR S0["Scanned image"] --> S1["OCR"] --> S2["Extracted text"] --> S3["Chunk and index"]
Key point: Images need OCR before anything else.
Why the other options are wrong
- A. A JSON parser
- A JSON parser needs JSON input.
- C. An HTML parser such as beautifulsoup
- HTML parsers read markup, which a scan does not have.
- D. A web crawler
- A crawler fetches web pages. It does not read images.