A Hybrid Approach and Unified Framework for Bibliographic Reference Extraction

Syed Tahseen Raza Rizvi, Andreas Dengel, Sheraz Ahmed

In: IEEE Access 8 Pages 217231-217245 IEEE 12/2020.


Publications are an integral part of a scientific community. Bibliographic reference extractionfrom scientific publication is a challenging task due to diversity in referencing styles and document layout.Existing methods perform sufficiently on one dataset however, applying these solutions to a different datasetproves to be challenging. Therefore, a generic solution was anticipated which could overcome the limitationsof the previous approaches. The contribution of this paper is three-fold. First, it presents a novel approachcalledDeepBiRDwhich is inspired by human visual perception and exploits layout features to identifyindividual references in a scientific publication. Second, we release a large dataset for image-based referencedetection with 2401 scans containing 38863 references, all manually annotated for individual reference.Third, we present a unified and highly configurable end-to-end automatic bibliographic reference extractionframework calledBRExSyswhich employsDeepBiRDalong with state-of-the-art text-based models to detectand visualize references from a bibliographic document. Our proposed approach pre-processes the imagesin which a hybrid representation is obtained by processing the given image using different computer visiontechniques. Then, it performs layout driven reference detection using Mask R-CNN on a given scientificpublication.DeepBiRDwas evaluated on two different datasets to demonstrate the generalization of thisapproach. The proposed system achieved an AP50 of 98.56% on our dataset.DeepBiRDsignificantlyoutperformed the current state-of-the-art approach on their dataset. Therefore, suggesting thatDeepBiRDis significantly superior in performance, generalized, and independent of any domain or referencing style.

Weitere Links

German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz