What is a research portal?

A portal reads one collection: the reports, papers, evaluations and guides an organisation has published or holds. It answers from that collection alone, so what people can ask about is what the organisation has loaded.

The original documents stay where they are. The portal reads a copy.

An answer is a short passage written from the documents. Each sentence that states a finding carries a numbered citation, and opening it shows the passage in the document, at the page used.

People need to know how far to trust an answer. In CorpusKit, figures, populations, named studies and years are tested against the cited text before the answer is shown. A sentence that fails is removed, and a note says so. Each answer is rated high, moderate or low confidence.

When the collection holds nothing that answers a question, the portal says so and shows the closest passages it found.

How it works

Search finds documents and passages by meaning, by exact words or both, and filters them by topic and document type. A library lists the whole collection, and suggested questions and topics give people a way in.

A portal can also hold the work of research: a knowledge graph of the names, places and topics in the collection and how they connect, investigations that keep a question open for months, briefings, comparisons and timelines written with references, and short knowledge checks.

Organisations with years of published research behind them:

• Research funders
• Government agencies
• Research institutes
• Evidence and policy institutes
• CRCs and research networks
• Knowledge hubs

Questions to ask before you choose one:

• Does every finding cite the page it came from, and can a reader open the document there?
• Are figures and years checked against the source before the answer is shown?
• What does the portal do when the collection holds nothing that answers a question?
• Where are the documents, their index and any knowledge graph held?
• Who can see the portal, and how do people sign in?
• Can it live at your own web address?
• How do new publications get in, and who loads them?
• Can you read the code, and run the portal yourself?

CorpusKit, the open source research portal for Progress Agentic RAG, is licensed under Apache 2.0. CorpusKit Cloud is Noice's hosted service for it: Noice runs the portal on your documents, applies upgrades and handles support.

corpuskit.org