SolutionTailor: Scientific paper recommendation based on fine-grained abstract analysis
Abstract: We developed a recommendation system that identifies papers proposing different solutions while sharing the same research objective. The system requires users to specify only the target research field and the abstract of a query paper. Using a neural language model, the system classifies abstract sentences into “background/objective’’ and “methodology’’ components, and performs recommendation based on the similarities derived from these components. Experimental results demonstrate that the system successfully recommends relevant literature beyond the existing citation network.
Authors: Tetsuya Takahashi
Publication venue: ECIR 2022 Demo
Demo
Reference
Tetsuya Takahashi and Marie Katsurai, “SolutionTailor: Scientific paper recommendation based on fine-grained abstract analysis,” Advances in Information Retrieval. ECIR 2022. Lecture Notes in Computer Science, vol 13186. Springer, Cham, 2022. PDF