The evidence supports an architectural claim: when an LLM forward pass is wrapped in an ontology-mediated auditing, structurally inevitable hallucination can be partially recovered.
Abstract
Large Language Models (LLMs) do not natively track the path of exclusions that a coherent discourse demands. When an input rests on a fabricated authority, a misapplied mechanism, or a surreptitious analogy, an unsteered LLM tends to engage with it as if it were well-posed, and this affects its generation. POLANYI++, an LLM-steering method that uses heuristics, ontologies and problem-solving methods for tacit-knowledge extraction, produces an Extended Knowledge Graph (XKG) in OWL2, but when a sophisticated nonsensical input is reified into the graph alongside the legitimate triples, it gets hardly detectable by automated reasoners, since the XKG is generated jointly with the wrong assumptions. We introduce DARKSIDE, a coherence-auditing method on top of POLANYI++. DARKSIDE formalises an explicit data structure of accumulated exclusions over discourse time, complemented by a warrant axis that classifies each named referent as Warranted, Unattested, Misattributed or Fabricated. The method is anchored in nine theoretical fragments unified under a shared deep frame of path integrity. The resulting DARKPOLANYI is evaluated as a steering layer over Gemini 3 on BSBench, a 100-item adversarial corpus of sophisticated-sounding nonsense across multiple domains, with Claude Sonnet 4.6 as an independent judge. DARKPOLANYI scores 1.89/2 mean versus 0.95/2 for the unsteered Gemini 3 Pro baseline; on the 97 cases with valid judgments in both arms, paired McNemar gives a paired bootstrap mean-diff = +0.92 (95% CI [+0.75, +1.08], p = 0.0001). The evidence supports an architectural claim: when an LLM forward pass is wrapped in an ontology-mediated auditing, structurally inevitable hallucination can be partially recovered. The XKG functions as the missing memory that LLMs lack, and the warrant axis as an epistemic firewall.
Large language models (LLMs) have transformed AI, yet they remain fundamentally limited by hallucination, unverifiable reasoning, and shallow evidence grounding. We argue that structure mining-rooted in decades of KDD research on taxonomy induction, ontology design, entity typing, and knowledge graph construction-is the key to overcoming these limitations. This tutorial presents a unified vision in which structuring serves as the enabling foundation for three pillars of next-generation LLM systems: (1) Structured Retrieval, where organizing corpora into ontology-guided multidimensional representations enables SQL-like queries that achieve substantially more precise and complete retrieval than similarity-based approaches; (2) Structured Reasoning, where grounding each inference step in typed, graph-structured evidence transforms opaque generation into auditable, verifiable reasoning chains; and (3) Structured Agent Memory, where multi-dimensional memory architectures bridge external corpus knowledge and experiential agent knowledge through a mutually enriching dual-memory design. Across all three pillars, we highlight how the cooperative interplay between classical KDD techniques and modern LLMs-where KDD defines structural schemas and quality constraints while LLMs execute flexible extraction and reasoning-creates systems that are more reliable, interpretable, and faithful. The tutorial covers both foundational methods and the latest advances (2024--2026), and concludes with open problems and future research directions at the intersection of data mining and LLMs.
Pengcheng Jiang, Jiashuo Sun, Wonbin Kweon et al.· Proceedings of the 32nd ACM...· 0 citations
This work presents the first systematic study of inverse relation directionality in LLMs, using a benchmark consisting of 5,457 instances spanning 27 distinct inverse relation labels and reveals systematic asymmetries in inverse relation classification across LLMs.
Safety alignment in multilingual models is uneven: a model that reliably refuses a harmful request in English will often comply with the same request in a lower-resource language. We trace this gap mechanistically in sarvam, an Indic-multilingual mixture-of-experts reasoning model, and find it is not a failure to detect harm. Harm is encoded as an internal direction that is nearly language-invariant in mid-network (English-vs-Indic cosine ${\approx}0.9$ at $L11$), and steering that direction upstream causally controls refusal. But the detection direction is orthogonal to the change that actually writes the refusal, which is late and assembled over the course of generation rather than read off in a single forward pass. We attribute the write to a specific, localizable circuit, a mixture-of-experts writer held in check by an attention opposer and price every way of intervening on it: damping the opposer is cheap and effective, amplifying the writer is a cost wall, and surgical edits to the responsible heads do nothing. The circuit's organization, and the gradient method that exposes it, recur in a second, unrelated MoE model, while the lever's strength is architecture-specific. The result is a cost-measured map of where a multilingual safety repair can land, and what it costs
Factual question answering (QA) typically assumes a single canonical answer, obscuring whether large language models (LLMs) retain divergent accounts of long-tail facts. To address this gap, we introduce ElephantBench, a closed-book knowledge probe comprising 1,094 questions generated through an auditable graph-based pipeline. The pipeline retrieves related documents from a low-exposure web corpus, identifies naturally occurring disagreements, and converts them into multi-account QA records. Each answer is verified against the originating documents and authoritative public web sources and is then reviewed by human annotators. Across 32 models, even the strongest model recovers both accounts on only 52.4% of questions, while on nearly all remaining questions it recalls one account but omits the other. Scaling model size and inference-time reasoning improve recall but do not eliminate this incompleteness. Corpus analysis further shows that exposure imbalance favors the dominant account, whereas greater minority-side exposure is associated with more complete recall. These findings establish ElephantBench as a reproducible knowledge probe for diagnosing epistemic myopia in parametric memory. More broadly, our graph-based benchmark construction pipeline provides an efficient and scalable way to turn long-tail corpora into source-traceable knowledge probes, supporting efforts to evaluate and advance the epistemic rigour of next-generation LLMs. Code is available at https://github.com/Tencent/ElephantBench.
Zhuo-Shi Pan, Jun-Ru Lu, Yan-Fei Qian et al.· 0 citations
SodaMem is presented, an evidence-grounded temporal graph memory that extracts typed FactEvents with mandatory provenance spans, persists mention time, occurrence time, and validity with SUPERSEDES/CONTRADICTS/UPDATES edges under hybrid lexical-dense indexing and answers via a planner-reader loop that gathers citable evidence before composing a final response.
Retrieval-Augmented Generation (RAG) is widely regarded as a novel paradigm born from the limitations of large language models (LLMs)--a mechanism to ground their outputs in external knowledge. This view, however, is incomplete when considered within a broader historical context. In this paper, we argue that the core ideas underlying RAG are not new: foundational concepts such as integrating retrieval and language generation, knowledge augmentation, answer verification, and iterative query (or prompt) refinement had already been studied and instantiated in information retrieval (IR) and question answering (QA) research dating back to the early 2000s, well before the emergence of LLMs. We make this case by systematically tracing the intellectual lineage of modern RAG and Agentic RAG back to their classical IR and QA antecedents, and examining why this continuity has gone under-recognized -- a consequence of community fragmentation, shifting terminology, and the recency bias endemic to fast-moving fields. Rather than treating LLMs as the origin point of retrieval-augmented intelligence, we propose viewing them as a new interface layer atop a decades-old QA architecture. This reframing is not merely historical: by situating RAG within the longer trajectory of IR research, we surface underutilized prior work -- on user modeling, answer validation, and query refinement -- that can directly inform next-generation RAG design, reducing unintentional rediscovery and fostering genuine cross-community integration.
Xiaoyan Zhao, Yujie Cai, Yang Zhang et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
