Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Leveraging the Power of Large Language Models in Entity Linking via Adaptive Routing and Targeted Reasoning

Yajie Li, Albert Galimov, Mitra Datta Ganapaneni, Pujitha Thejaswi, De Meng, Priyanshu Kumar, Saloni Potdar

TL;DR

ARTER tackles the high cost of LLM-based reasoning in entity linking by introducing adaptive routing that separates easy and hard mentions. It leverages a lightweight router over signals including $\theta_1(i)$, $\theta_2(i)$, $\theta_3(i)$, and $\phi(i)$ to route hard cases to a targeted LLM prompt while resolving easy cases with a fast linker like ReFinED. The approach achieves up to +4.47% accuracy gains over ReFinED and remains competitive with full prompting, while reducing LLM token usage by around 58% on average. This delivers practical deployment benefits by avoiding domain-specific fine-tuning and controlling latency and cost, especially under domain shifts and knowledge-base updates.

Abstract

Entity Linking (EL) has traditionally relied on large annotated datasets and extensive model fine-tuning. While recent few-shot methods leverage large language models (LLMs) through prompting to reduce training requirements, they often suffer from inefficiencies due to expensive LLM-based reasoning. ARTER (Adaptive Routing and Targeted Entity Reasoning) presents a structured pipeline that achieves high performance without deep fine-tuning by strategically combining candidate generation, context-based scoring, adaptive routing, and selective reasoning. ARTER computes a small set of complementary signals(both embedding and LLM-based) over the retrieved candidates to categorize contextual mentions into easy and hard cases. The cases are then handled by a low-computational entity linker (e.g. ReFinED) and more expensive targeted LLM-based reasoning respectively. On standard benchmarks, ARTER outperforms ReFinED by up to +4.47%, with an average gain of +2.53% on 5 out of 6 datasets, and performs comparably to pipelines using LLM-based reasoning for all mentions, while being as twice as efficient in terms of the number of LLM tokens.

Leveraging the Power of Large Language Models in Entity Linking via Adaptive Routing and Targeted Reasoning

TL;DR

ARTER tackles the high cost of LLM-based reasoning in entity linking by introducing adaptive routing that separates easy and hard mentions. It leverages a lightweight router over signals including , , , and to route hard cases to a targeted LLM prompt while resolving easy cases with a fast linker like ReFinED. The approach achieves up to +4.47% accuracy gains over ReFinED and remains competitive with full prompting, while reducing LLM token usage by around 58% on average. This delivers practical deployment benefits by avoiding domain-specific fine-tuning and controlling latency and cost, especially under domain shifts and knowledge-base updates.

Abstract

Entity Linking (EL) has traditionally relied on large annotated datasets and extensive model fine-tuning. While recent few-shot methods leverage large language models (LLMs) through prompting to reduce training requirements, they often suffer from inefficiencies due to expensive LLM-based reasoning. ARTER (Adaptive Routing and Targeted Entity Reasoning) presents a structured pipeline that achieves high performance without deep fine-tuning by strategically combining candidate generation, context-based scoring, adaptive routing, and selective reasoning. ARTER computes a small set of complementary signals(both embedding and LLM-based) over the retrieved candidates to categorize contextual mentions into easy and hard cases. The cases are then handled by a low-computational entity linker (e.g. ReFinED) and more expensive targeted LLM-based reasoning respectively. On standard benchmarks, ARTER outperforms ReFinED by up to +4.47%, with an average gain of +2.53% on 5 out of 6 datasets, and performs comparably to pipelines using LLM-based reasoning for all mentions, while being as twice as efficient in terms of the number of LLM tokens.
Paper Structure (28 sections, 5 equations, 2 figures, 11 tables)

This paper contains 28 sections, 5 equations, 2 figures, 11 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Method
  4. Problem Formulation
  5. Candidate Generation
  6. Router
  7. Reasoning Module
  8. Prompting Strategies
  9. Experimental setup
  10. Datasets
  11. Data source
  12. Entity Linking Evaluation Metric
  13. Entity Linking Baselines
  14. Results
  15. Router Performance
  16. ...and 13 more sections

Figures (2)

  • Figure 1: ARTER leverages a light-weight router on top of the given context and retrieved candidate entities to categorize contextual mentions into easy and hard cases. The cases are then handled by a low-computational entity linker (e.g. ReFinED) and more expensive targeted LLM-based reasoning respectively.
  • Figure 2: Entity linking accuracy vs. total inference cost across LLM-based configurations.