Original title: Improvement of personalized recommendation service level of intelligent libraries based on artificial intelligence and big data mining technology
Chao Zhang · September 2026
College libraries want to recommend books students actually like. Old systems struggle because data is scarce and student tastes change instead of staying frozen. Researchers built a new tool called SVD-LGCN to fix this. It uses a graph network to connect students and the books they read. This lets the system track complex reading patterns across the whole campus. At the same time, it blends in an idea from a math method called SVD++. That technique collects everything a reader has checked out over time to build a lively, moving picture of their interests. The researchers tested their system on a public collection of book ratings called Book-Crossing. They found that it beat a standard setup called LightGCN, lifting a key ranking test score by 4.1%. It gives college libraries a better way to suggest books.
Imagine you're at a used bookstore and the clerk asks what you like. You tell them sci-fi, and they bring you a stack of sci-fi books. But after you browse a few, they notice you keep passing over space operas and spending time on near-future thrillers. So they swap the stack for thrillers. Most recommendation systems only ask once and never adjust. Here, the system watches your every interaction and updates the stack accordingly, using all the books you've handled to refine its guess.
When a university library's recommendation system learns your taste only once, it misses what you actually want next. This paper shows a way to update the recommendation based on every book you've interacted with, making suggestions more accurate. For students and researchers, that means less time searching and more time discovering useful books.
The abstract only tests on one public dataset (Book-Crossing), so generalizability to other library systems or data types is not established. It reports improvements over a single baseline (LightGCN) without comparing to other state-of-the-art recommendation models. The method's computational cost and scalability for very large collections are not discussed.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Improvement of personalized recommendation service level of intelligent libraries based on artificial intelligence and big data mining technology — Chao Zhang — Scientific Reports — doi:10.1038/s41598-026-72513-y
Abstract licence: CC BY
Explanation generated by Synthesis from the abstract — it is not the original paper.