Peilan Xu
Researcher
Nanjing University of Information Science and Technology
Peilan Xu is a researcher at Nanjing University of Information Science and Technology. Xu is corresponding author on FeatGEO, the ACL 2026 paper that reframes generative citation optimization as feature-level multi-objective control over structure, content, and language.
Corresponding Author on FeatGEO
Xu is the corresponding author on FeatGEO, co-authored with Zikang Liu. The paper's claim is that generative citation work should operate at feature level, not only through token-level rewrite tricks.
Optimizing Webpage Features
FeatGEO abstracts webpages into interpretable structural, content, and linguistic properties, then optimizes over that feature space before realizing configurations in natural language. The goal is control over the trade-off between citation visibility and content quality.
Experiments on GEO-Bench across three generative engines show consistent citation-visibility improvements while maintaining or improving content quality, outperforming token-level baselines. Analyses in the paper find citation behavior more strongly influenced by document-level content properties than by isolated lexical edits.