Taxonomy Maintenance In The Wild Over Evolving Scholarly Data: Reliability, Efficiency, and Cost-Effectiveness
2026-07-10 • Databases
Databases
AI summaryⓘ
The authors address the problem of keeping scientific topic trees (taxonomies) up-to-date as new research papers appear. They propose a system called GIST that builds these trees by using expert-verified clues from paper sections to piece together parts of a hierarchy. GIST then combines these parts in a special geometric space to understand relationships and updates efficiently without starting from scratch. Their tests on real paper collections show that GIST is more accurate and much faster than previous methods.
taxonomy maintenancearXivhierarchical clusteringbox embeddingslanguage modelsrelated work sectionincremental learningnovelty detectionword embeddingsevidence retrieval
Authors
Daomin Ji, Hui Luo, Zhifeng Bao, Junhao Gan, Zi Huang
Abstract
The rapid growth of scientific publications makes scholarly taxonomies quickly obsolete. We study taxonomy maintenance in the wild, a new problem that moves beyond static construction by continuously adapting taxonomies to evolving scholarly repositories, such as arXiv, for a given research topic. We propose GIST, a robust framework for maintaining evolving taxonomies. Unlike purely LLM-centric approaches, GIST grounds structure induction in expert-curated evidence by extracting partial hierarchies from the "Related Work" sections of papers. It integrates these partial taxonomies into a unified global taxonomy in a geometric box-embedding space, where box containment encodes the inductive bias of is-a relations. To connect semantics with geometric structure, GIST learns a bidirectional mapping between word embeddings and box embeddings. For efficient incremental updates, GIST uses novelty-aware coreset selection to update the model with representative historical signals and new evidence, avoiding costly full retraining. To handle high-velocity paper streams under user-specific token budgets, GIST further combines a hypothesized concept generator with a cost-effective evidence retrieval module. Experiments on real-world arXiv datasets show that GIST outperforms state-of-the-art baselines, improving Node F1 and Edge F1 by 11.0% and 13.1% over the strongest baseline while requiring only 9.6% of its runtime and 12.7% of its monetary cost.