PolitNuggets: Benchmarking Agentic Discovery of Long-Tail Political Facts

2026-05-13

AI Summary

Existing agent systems lack effective evaluation frameworks for discovering and synthesizing fine-grained, multilingual long-tail political facts from decentralized sources. To address this gap, this work introduces PolitNuggets, a multilingual benchmark comprising over 10,000 political facts pertaining to 400 global political figures, and proposes FactNet, an evidence-conditioned protocol that standardizes the assessment of agents across three dimensions: fact discovery, fine-grained accuracy, and efficiency. Leveraging an optimized multi-agent architecture integrated with cross-lingual information extraction and tool invocation mechanisms, our experiments reveal pervasive challenges in current systems, particularly in fine-grained reasoning and computational efficiency. Key bottlenecks include limited short-context extraction capability, insufficient multilingual robustness, and unreliable tool utilization.

Abstract

Large Reasoning Models (LRMs) embedded in agentic frameworks have transformed information retrieval from static, long context question answering into open-ended exploration. Yet real world use requires models to discover and synthesize "long-tail" facts from dispersed sources, a capability that remains under-evaluated. We introduce PolitNuggets, a multilingual benchmark for agentic information synthesis via constructing political biographies for 400 global elites, covering over 10000 political facts. We standardize evaluation with an optimized multi agent system and propose FactNet, an evidence conditional protocol that scores discovery, fine-grained accuracy, and efficiency. Across models and settings, we find that current systems often struggle with fine-grained details, and vary substantially in efficiency. Finally, using benchmark diagnostics, we relate agent performance to underlying model capabilities, highlighting the importance of short-context extraction, multilingual robustness, and reliable tool use.

Problem

• long-tail facts
• agentic information synthesis
• political biographies
• multilingual benchmark
• fact discovery

Innovation

• agentic information synthesis
• long-tail facts
• multilingual benchmark
• FactNet
• Large Reasoning Models