So, you’re constructing a RAG system or using an LLM to speak with documents. But users often ask: how can we trust the answers?
Furthermore, we steadily hear about hallucinations, which undermine users’ trust.
If we construct an application but fail to indicate users where the answers come from, the applying might turn out to be unusable in some cases.
In this text, I’ll share an approach to handle this concern. By linking every answer generated by the LLM to its source text within the document, we will construct transparency and trust. This method not only provides clear evidence for the answers but additionally allows users to confirm the outcomes directly inside the PDF.
Sometimes, the generated answer is probably not perfectly accurate, but having the ability to locate the right source text is already helpful for the user.
Let’s take an example of this paper from arxiv.org. We will imagine this use case:
Step one on this approach is to extract the text from the PDF in a structured format.
