6arXiv cs.CL (Computation and Language)·
Bridge Evidence: Static Retrieval Utility Does Not Predict Causal Utility in Multi-Step Agentic Search
A new arXiv paper demonstrates empirically that static retrieval relevance scores are nearly statistically independent from causal utility in multi-step agentic search (Spearman rho = -0.026 across 23,322 document observations). Using a ReAct-style agent over HotpotQA with counterfactual trajectory replays, the authors show that roughly a third of documents are causally load-bearing ('bridge documents') while appearing useless to a static reader. The mechanism is identified: bridge documents supply discriminative entities that redirect subsequent queries, meaning optimizing static RAG utility does not deliver agentic utility.
Related guides (2)
Related events (8)
Anthropic introduces Contextual Retrieval to reduce RAG retrieval failures by up to 67%
Anthropic published a technical method called Contextual Retrieval that combines Contextual Embeddings and Contextual BM25 to address the context-loss problem in traditional RAG pipelines. The approach prepends chunk-level context before encoding, reducing failed retrievals by 49% standalone and 67% when combined with reranking. The post also highlights prompt caching as a simpler alternative for knowledge bases under 200K tokens, and provides a cookbook for deployment with Claude.
Comparative Study: Semantic Metadata vs. Unstructured Web Retrieval for Agentic Data Discovery
This paper evaluates whether LLM-based agents still need structured semantic metadata (e.g., schema.org) for data retrieval, comparing a Baseline Agent searching open-web documents against a Semantic Agent leveraging 90 million schema.org-annotated datasets. Using an LLM-as-a-judge pipeline aligned to FAIR principles, the Semantic Agent achieves 65.7% higher overall precision in retrieving FAIR-compliant datasets, while the Baseline Agent answers 40% more questions but frequently returns prose-heavy or portal landing pages instead of actionable data. The study concludes that structured semantic ecosystems remain essential for reliable, execution-oriented agentic workflows despite LLMs' broad unstructured retrieval capabilities.
Coverage Illusion: Post-Retrieval Cascade Design Reduces LLM Augmentation Overhead in Production RAG
A case study on the Danish National Encyclopedia's RAG system evaluates five retrieval workflows across 20,000 query-workflow pairs, revealing a 'Coverage Illusion' where synthetic queries overestimate the need for LLM augmentation (90%+) versus real production traffic (27.8%). Pre-retrieval routing cannot detect this gap because augmentation necessity is only revealed after index search. A post-retrieval cascade running workflows cheapest-first and escalating to LLM augmentation only on empty results improves quality by +0.140 Composite Overall points over Always-HyDE, reduces latency by 31.8%, and eliminates LLM augmentation for 72.2% of real queries. The work highlights a structural mismatch between synthetic and real query distributions that affects RAG system design assumptions.
HKVM-RAG: Hypergraph key-value separation improves multi-hop retrieval-augmented generation
A new arXiv preprint introduces HKVM-RAG, an evidence-organization layer for multi-hop RAG that uses weighted hyperedges as retrieval keys while retaining passage text as answer values. Under a fixed-substrate protocol controlling for tuple cache, reader, and evaluation budget, the hypergraph key-value approach improves over KG-PPR by +3.4 F1 on 2WikiMultiHopQA and +3.6 F1 on MuSiQue. A dense-aware controller combining frozen ColBERTv2 with HKVM features reaches 88.8, 65.1, and 85.8 F1 on three benchmarks, outperforming ColBERTv2 alone by 5–11 F1 points. The work positions hypergraph organization as a reusable evidence-control mechanism rather than a dense-retrieval replacement.
Causal audit of neuron selectors reveals refusal behavior lives in redundant subspace
A new arXiv preprint introduces a causal auditing framework for neuron-row attribution methods in LLMs, using one-shot zeroing interventions to test whether high-scoring neurons are genuinely causally important. The authors find that attribution methods outperform activation/magnitude baselines for identifying dispensable rows, and that refusal behavior (for hate/crime prompts) can be installed or removed via attributed rows while preserving fluency. A key finding is that refusal occupies a redundant subspace — different attribution methods identify largely disjoint sufficient row sets — meaning no single method recovers a unique causal mechanism. Rank-stability, a common proxy for selector quality, is shown to be a poor indicator of causal validity.
AGORA benchmark tests agentic document reasoning over large authentic workplace archives
Researchers introduce AGORA, a benchmark pairing 362 questions with 9,664 authentic workplace documents (372M tokens across eight domain collections) to evaluate archive-grounded agentic reasoning. The benchmark is designed so documents far exceed any model's context window, forcing deliberate exploration rather than exhaustive scanning. Evaluating eight models, the best achieves only 59.4% accuracy, indicating the task is far from solved. The benchmark addresses a gap in existing evals that do not jointly stress archive-groundedness, agentic exploration, and cross-domain coverage.
Probe Trajectories Reveal Reasoning Dynamics in Large Reasoning Models
This paper investigates whether hidden representations of Large Reasoning Models (LRMs) can predict future model behavior by analyzing probe trajectories—the continuous evolution of concept probabilities across Chain-of-Thought reasoning tokens. The authors find that temporal trajectory features (volatility, trend, steady-state) significantly outperform single static probes, with max-pooling achieving up to 95% AUROC across safety and mathematics domains. Two methodological insights are offered: template-based training data matches dynamically generated responses in quality, and pooling strategy is critical to probe performance. The work positions probe trajectories as a complementary safety monitoring framework for LRMs where CoT faithfulness cannot be assumed.
Study compares human and LLM active causal reasoning, finding LLMs less efficient but near human-level on conjunctive rules
A new arXiv paper investigates whether active exploration reduces the 'conjunctive handicap' in causal learning, using a blicket detector task with adult participants who could freely intervene to identify causal objects. Results show active exploration substantially improves human conjunctive causal reasoning, though conjunctive rules still require more tests than disjunctive ones. State-of-the-art LLMs approach human-level hypothesis inference accuracy but show less efficient exploration strategies and similar conjunctive-disjunctive performance gaps, raising questions about the nature of LLM causal reasoning.