Search to context

A common pattern for RAG and agents is to search, then expand each match with the content around it. You give the model a few relevant passages instead of whole papers.

Build context from search results

const question = "How did revenue change in Europe?";

const { items } = await paperful.search.query(question);
const [result] = items;

const graph = await paperful.papers.graph(result.paper.id);
const node = graph.nodes.find((n) => n.id === result.nodes[0].id);

const context = await node?.context({ before: 3, after: 3 });

Cite the source

Every node keeps its page and bounding box, so you can show users where an answer came from.

for (const { id, page } of result.nodes) {
  console.log(`${result.paper.fileName}, page ${page}`);
}

Paperfuldocs

Build context from search results

Cite the source

On this page