References

Every answer ResponseHub produces is built from references — the specific material it drew on. Each reference is one of two things:

• A knowledge base item — an approved answer you already have
• A chunk of a source — a passage of text from one of your documents

References are what make an answer checkable. Rather than trusting the answer on its own, you can look at what produced it.

How sources become chunks

A policy document is too long to work with whole, so each source is split into chunks when it is imported. A chunk is roughly 200 words — around 900 characters — cut at sentence boundaries so it never splits mid-sentence, with a little overlap between neighbouring chunks so a passage that spans a boundary isn’t lost.

Every chunk remembers where it came from: which document, and which section of that document. That is why a reference can point you to the exact part of a policy behind an answer.

How references are used

Sometimes a single chunk contains the answer outright. More often the answer is inferred across several — one chunk establishing that you encrypt data, another naming the standard, a third covering key management.

That is why the references list usually holds more than one entry, and why they are worth reading together rather than one at a time.

Reading the references card

Each reference is badged:

Badge	Means
Used in answer	The material informed the answer
Used as answer	The text was reused word for word — see verbatim answers

Anything retrieved but not drawn on appears separately under Unused references. Each reference also carries a relevance rating.

When there are no references

If nothing in your material matches the question closely enough, the card shows Insufficient Context and explains what was missing rather than showing an answer. That means the topic isn’t covered yet — either import a source that covers it, or write the answer yourself.

Next steps

• Relevancy
• Confidence