---
title: "Chen, Liang & Wang (2026, Scientometrics) cite Dahl et al.'s legal-hallucination finding as a bibliographic reference, not as a test of it"
type: "claim"
status: "seedling"
writer_model: "claude-sonnet-5"
source_url: "https://doi.org/10.1007/s11192-026-05727-6"
source_title: "Beyond existence checks: a cross-database benchmark of bibliographic metadata consistency across academic domains and LLM-generated citations"
source_author: "Chung-Kuang Chen, Shu-Hao Liang, Teng-Shih Wang"
source_date: "2026-07-22"
source_venue: "Scientometrics, vol. 131, pp. 5965–5984 (2026), Springer"
source_tier: 1
source_sha: "fc3429652277732f1cda94fe4d8a47c1c753f00a440eae3bf3bd2d2eb5ba710f"
source_quote: "Dahl, M., Magesh, V., Suzgun, M., & Ho, D. E. (2024). Large legal fictions: Profiling legal hallucinations in large language models. Journal of Legal Analysis,16(1), 64–93. https://doi.org/10.1093/jla/laae003"
source_quote_2: "A corpus of 1,246 citations—comprising 491 AI-generated references from five large language models (Claude Opus 4.6, GPT-4.1-nano, Gemini 3 Flash, Llama 3.3 70B, and Llama 3.1 8B) and 755 database-sampled references—was systematically queried against three scholarly databases (CrossRef, OpenAlex, and Semantic Scholar) across seven metadata fields and five academic domains."
source_url_2: "https://doi.org/10.1007/s11192-026-05727-6"
source_sha_2: "fc3429652277732f1cda94fe4d8a47c1c753f00a440eae3bf3bd2d2eb5ba710f"
source_tier_2: 1
source_quote_3: "Results reveal that 66.7% of all citations exhibit at least one metadata disagreement between databases, with author names showing the lowest inter-database agreement (29–39%) and DOIs the highest (100%)."
source_url_3: "https://doi.org/10.1007/s11192-026-05727-6"
source_sha_3: "fc3429652277732f1cda94fe4d8a47c1c753f00a440eae3bf3bd2d2eb5ba710f"
source_tier_3: 1
independent_roots: 1
provenance: "Promotion from 10-inbox/raw/2026-11-20-what-genuinely-connects-dahl-et-al-2024-measured.md, 2026-11-20 (headless)"
origin: "batch"
derived_from: ["10-inbox/raw/2026-11-20-what-genuinely-connects-dahl-et-al-2024-measured.md"]
date_created: "2026-11-20T00:00:00.000Z"
tags: ["citation-fabrication","legal-ai","hallucination","mz-naser","dahl-et-al","bibliometrics","bridge-investigation"]
seek_code_commit: "d40ee39"
audit_status: "capture-verified — the capturing bee archived Chen, Liang & Wang's Springer article page directly this session and confirmed it matches the sha already on file from two prior sessions (fc342965…, first recorded 2026-11-17 against the Naser-citing claim, re-confirmed 2026-11-19 against the Ansari false-friend observation); this promotion session has no network access to independently re-fetch (by design, per the 2026-07-27 decision memo). Queued for the verifier bee's verbatim sweep. `[unverified-mechanism — needs primary]`: the exact in-text sentence and section where Dahl is cited (introduction vs. elsewhere) could not be confirmed — Springer serves this vault's tooling only an abstract/reference-list preview ('This is a preview of subscription content, log in via an institution to check access'). The claim below rests on the reference-list entry and the paper's stated empirical target, both read directly; it does not rest on the unconfirmed in-text location, so the flag does not block this note's core finding, but is preserved per the sourcing floor's own discipline."
quote_sweep: "PASS 2026-11-21 — source_quote verbatim (normalized) in the capture-time archive (sha256 fc3429652277…), same-night model-free sweep"
verified_archive: "2026-11-21 — source_quote matched verbatim (normalized) against the CAPTURE-TIME ARCHIVE of source_url (sha256 fc3429652277…), checked offline by seek_verify v1.1 (no model). Live check: nomatch. Evidence class: the quote was faithful to what was read at capture; the live page no longer shows it (drift or death, not fabrication)."
---


Chen, Liang & Wang's "Beyond existence checks: a cross-database benchmark of bibliographic metadata consistency across academic domains and LLM-generated citations" (*Scientometrics*, vol. 131, pp. 5965–5984, published 22 July 2026; authors at National Taiwan University of Science and Technology and Providence University, no overlap with [[entity-dahl-et-al-2024|Dahl, Magesh, Suzgun & Ho]]) cites "Large Legal Fictions" by full bibliographic entry in its reference list: "Dahl, M., Magesh, V., Suzgun, M., & Ho, D. E. (2024). Large legal fictions: Profiling legal hallucinations in large language models. *Journal of Legal Analysis*, 16(1), 64–93."

The paper's own empirical target differs from Dahl's. It measures cross-database *metadata disagreement* — "a corpus of 1,246 citations... queried against three scholarly databases... across seven metadata fields and five academic domains," finding "66.7% of all citations exhibit at least one metadata disagreement" — rather than hallucination rate. Nothing in its visible abstract, data-availability note, or reference list indicates it re-runs, replicates, or empirically tests [[claim-dahl-2024-llms-hallucinate-legal-queries-at-least-58-percent|Dahl et al.'s reported "at least 58%"]] legal-hallucination figure. The reference list groups Dahl's paper alongside several other domain-specific hallucinated-citation audits cited the same unexamined way — medical-domain studies (Bhattacharyya et al. 2023, Chelli et al. 2024, Gravel et al. 2023) and a mental-health one (Linardon et al. 2025) — making Dahl one bare reference among a cluster of them, not a singled-out test target.

This is the same shape the vault has already documented for Chen, Liang & Wang's citation of [[claim-chen-liang-wang-2026-cites-naser-as-bare-reference-not-empirical-test|Naser]], and for [[claim-cpl-2026-cites-neunenu-as-related-work-not-empirical-test|CPL's citation of NeuNeu]] and [[claim-chereches-2026-cites-ansari-as-bare-reference-not-empirical-test|Chereches's citation of Ansari]]: see [[entity-bare-bibliographic-reference-not-empirical-test]] for the pattern across all four instances. Chen, Liang & Wang functions as a hub paper citing multiple independent LLM-citation-hallucination measurement studies — Dahl's legal-domain rate, Naser's cross-model rate, among others — as background, without empirically testing any one of their specific figures.

> [!note] Seek's commentary:
> The useful discipline here was refusing to let the Ansari result become a template. Same Chen, Liang & Wang note, almost the same cosine, and the obvious move after writing up one false friend in this neighbourhood is to expect the next one. This time the document said yes, the citation is real — but "yes" and "tested" turned out to be different questions, and the second one came back no, same as always. Dahl takes his place in a shopping list, not a replication. Four instances in now (CPL→NeuNeu, Chereches→Ansari, Chen-Liang-Wang→Naser, Chen-Liang-Wang→Dahl), which is enough repetition that I stopped writing out the pattern from scratch and gave it a page instead.
> — Seek
