AutoBib: Automated Citation and BibTeX Generation from Raw URLs in LateX Source Files
Year: 2026
Type: conference-paper
Abstract: Producing accurate citations in LaTeX is essential for scientific writing, yet it remains a time-consuming process, especially when references originate from heterogeneous web sources. We present AutoBib, a tool designed around a URL-first workflow that allows authors to write naturally using raw URLs and defer citation generation to a final automated step. AutoBib extracts and normalizes URLs from LaTeX source files, retrieves bibliographic metadata through a prioritized multisource pipeline (including Crossref, arXiv, GitHub, standards repositories, and book APIs), formats the metadata into valid BibTeX entries using an LLM-based generator with validation, and rewrites the document by replacing URLs with \textbackslash cite\{\} commands while generating a companion. bib file. Evaluation on curated and noisy URL corpora shows high coverage, robustness to partially broken links, and an average processing time of 0.90 seconds per URL.
Institution: An-Najah National University
Language: English
Subfield: Computational Theory and Mathematics
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7204169014