# Geodesic Semantic Search: Cartographic Navigation of Citation Graphs with Learned Local Riemannian Maps

Source: [Geodesic Semantic Search: Cartographic Navigation of Citation Graphs with Learned Local Riemannian Maps](<https://www.sota2.com/research/paper/69a519582d603e6de3e3c6d9>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-06-06T05:00:11.965Z

Authors: Brandon Yee, Lucas Wang, Kundana Kommini

Published: 2026-02-27T04:17:41.000Z

## Abstract

We present Geodesic Semantic Search (GSS), a retrieval system that learns node-specific Riemannian metrics on citation graphs to enable geometry-aware semantic search. Unlike standard embedding-based retrieval that relies on fixed Euclidean distances, \\gss{} learns a low-rank metric tensor $\\mL\_i \\in \\R^{d \\times r}$ at each node, inducing a local positive semi-definite metric $\\mG\_i = \\mL\_i \\mL\_i^\\top + \\eps \\mI$. This parameterization guarantees valid metrics while keeping the model tractable. Retrieval proceeds via multi-source Dijkstra on the learned geodesic distances, followed by Maximal Marginal Relevance reranking and path coherence filtering. On citation prediction benchmarks with 169K arXiv papers, GSS achieves 23\\% relative improvement in Recall@20 over SPECTER+FAISS baselines. We provide a Bridge Recovery Guarantee characterizing when geodesic retrieval qualitatively outperforms direct similarity, a margin separation result connecting training loss to retrieval quality, and characterize the expressiveness of low-rank metric parameterization. Our hierarchical coarse-to-fine search with k-means pooling reduces computational cost by $4\\times$ while maintaining 97\\% retrieval quality.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Citation Prediction on arXiv citation network 169K papers (2022+)](<https://www.sota2.com/research/sota/citation-prediction-on-arxiv-citation-network-169k-papers-2022-plus.md>): Citation prediction; arXiv citation network 169K papers (2022+); R@10: 39.8.

- [Concept Bridging on arXiv](<https://www.sota2.com/research/sota/concept-bridging-on-arxiv.md>): Concept Bridging; arXiv; Bridge@10: 45.6.

- [Semantic Search on arXiv](<https://www.sota2.com/research/sota/semantic-search-on-arxiv.md>): Semantic Search; arXiv; nDCG@10: 61.2.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2602.23665>)

[PDF](<https://arxiv.org/pdf/2602.23665.pdf>)
