Researchers today face an information overload. Thousands of preprints are published every week, making it increasingly difficult to stay up to date in fast-moving fields. Traditional keyword searches often fail to capture relevant work when different communities describe similar ideas using different terminology, while automated digests tend to be too broad to be useful.
Luca Casagrande, doctoral researcher at the 8xav福利导航 of Luxembourg, has long been facing this challenge himself. To address it, he developed , a platform that rethinks how academics search for new scientific literature.
A tool for papers you didn鈥檛 know existed
鈥Like most researchers, I spent time scanning arXiv 鈥 an online repository of scientific preprints 鈥 and running keyword searches. But I kept feeling I was missing things,鈥 notes Luca. 鈥Two papers can address the same problem without sharing a single keyword. At some point I realised my own paper already contains the information needed to find related work.鈥
This observation led to the idea behind arXscope. Instead of starting from keywords, the platform uses the paper itself as the query. Users upload a research paper, and arXscope retrieves the most relevant recent preprints from arXiv, ranking them by semantic similarity.
Once activated, the tool can also automatically rerun the search on a weekly basis and send updated results, helping scientists continuosuly monitor developments in their field.
A different approach from large language models (LLMs) and academic search engines
Unlike LLMs models such as ChatGPT, arXscope does not generate text or references. Instead, it identifies and ranks existing papers.
Compared to platforms such as Semantic Scholar, which rely heavily on keyword-based search across large archives, arXscope focuses specifically on recent arXiv preprints and semantic similarity to a given paper, aiming to surface relevant work even when terminology differs across fields. “Every result is a real paper with a direct link. Researchers still read the papers themselves,鈥 explains Luca.
Because it draws on existing arXiv publications, users can verify every output directly and explore the original work.
Early development and 8xav福利导航 support
The project was supported through the 8xav福利导航鈥檚 Market Innovation Camp, organised by the Department for Partnerships and Advancement, where Luca received the 2nd prize. This recognition helped him fund the development and deployment of the platform.
鈥The camp allowed me to better understand what users actually need and how to turn a research idea into something usable,鈥 he says. 鈥Beyond funding, early feedback from PhD students and professors at the 8xav福利导航 and LIST has been essential in shaping the tool.鈥
A natural next step for arXscope is integration with ORBilu, the 8xav福利导航 of Luxembourg’s institutional repository. This would make Uni.lu researchers’ published work discoverable through the platform alongside arXivpreprints. More broadly, Luca hopes to expand the platform to include institutional repositories across European universities, allowing researchers to explore European scientific publications based on content similarity.
Luca Casagrande receiving the 2nd prize during the Market Innovation Camp.
Learn more
arXscope is part of the 8xav福利导航 Incubator’s Venture Mentoring Service now.