Mission accomplished: Apollo AI search solution launched

Apollo holds the University’s open access research output, which attracts over 260,000 visits per month from across the globe, and has delivered around ten million downloads in the past five years. iKVA was tasked with creating a bespoke search solution that would enable Apollo users to discover contextually relevant information quickly and easily, without needing to construct complex keyword searches, while bypassing traditional channels such as research teams. 

Apollo users can now choose to upload a compatible full text object, such as a PDF or Word document, a link to a text-based website or a typed text-based search to the Apollo website to access research papers, journal articles and other academic knowledge. Individual item pages will also offer recommendations based on iKVA's analysis, and the research results are delivered without interrupting the user’s workflow.

Jon Horden, CEO, iKVA, said: “Our underlying technology is based on innovative research from academics at Cambridge University’s Computer Laboratory and being able to use that research to support the University Library team has been particularly rewarding. Deploying advanced vector mapping technology, combined with neural networks, improves search accuracy and relevancy, based on the full context of a document, compared with more traditional NLP (Natural Language Processing) and keyword-based techniques. We are confident that this will provide easier access to the readily available research from one of the world’s leading Universities and encourage more people to use Apollo.”

Agustina Martinez-Garcia, Open Research Systems Manager, Cambridge University Libraries, said: “Making the University of Cambridge’s research available to the outside world is a core tenet of the Apollo repository and we are always looking for new ways to help people discover the content on Apollo. This solution from iKVA offers users new search and discovery opportunities via a tool that integrates into our existing repository interface. It’s been a pleasure to work with the team at iKVA. They were quick throughout to understand our requirements and very helpful during implementation.”

The new search service is available here: https://www.repository.cam.ac.uk/page/ikva-search.

Image: Agustina Martinez-Garcia, Open Systems Research Manager, The University of Cambridge and Jon Horden, CEO, iKVA