cs.CLAug 27, 2026

Pair-Level Essay-Scale Republication and Reuse from Fragmented Historical Text Reuse: A Workflow Study on Eighteenth-Century Books and Newspapers

Authors: Ke Shu, Kira Hinderks, Eetu Mäkelä, Mikko Tolonen

Organizations: University of Helsinki Helsinki, Finland

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

This paper addresses the recovery of essay-scale republication and reuse from fragmented text-reuse evidence, a setting whose central challenge is pair-level evidence consolidation and not fragment retrieval alone. The study focuses on a candidate set centered on essays by eighteenth-century Scottish philosopher David Hume, spanning books from ECCO (Eighteenth Century Collections Online) and historical newspapers. Because the input consists of fragmented reuse hits instead of clean document pairs, and positive coverage is inherently incomplete, we formulate the task as pair-level evidence consolidation into plausible transmission relations and compare three methodological families: a staged rule-based workflow, baselines (a decision tree and two direct LLM settings), and automated rule adaptation. On labeled ECCO--ECCO slices, pair-level feature aggregation alone already reaches 0.948 F1 on the main labeled slice, while the final workflow gives the strongest overall precision-recall trade-off among the tested rule stages. On the full ECCO--ECCO candidate universe, direct LLM baselines flag up to 14,886 pairs as reprints compared to 771 for the final workflow, behaving in this direct-prompt setup as high-recall candidate expanders rather than precision-controlled deployment classifiers. On ECCO--Newspaper, manual audit confirms all 176 predicted positives as genuine cases of republication or reuse, while issue duplication and source-side multiplicity reveal additional provenance structure. Under incomplete ground truth, auditable pair-level evidence consolidation provides a practical way to produce compact candidate spaces for historical inspection.

Figures & tables

Feature groupDescription
CoverageFraction of source essay spanned by reuse hits
Span / bundleMax and cumulative span of contiguous hit bundles
Fragment chainLongest monotone hit chain ordered by source offset
Section conc.Hit concentration across document sections
Cue signalsTitle, heading, quotation, paratext hit counts
Table 1. Pair-level features used across all methods.
Figure 1. Overview pipeline: fragment hits are aggregated into pair-level features, then classified by the staged rule workflow.
GateRule logic
StructuralAccept compact chains with span ≥\geq 900, coverage ≥\geq 0.09, sum coverage ≥\geq 0.10, section count ≤\leq 8, fanout ≤\leq 40, and chain score ≥\geq 1200; or denser bundles with span ≥\geq 650, coverage/sum coverage ≥\geq 0.15, section count ≤\leq 4, fanout ≤\leq 8, and at least two fragments.
CoverageAlso accept high-coverage pairs with coverage ≥\geq 0.30, fanout ≤\leq 40, and section count ≤\leq 8.
ContextRequire either title/heading evidence or the absence of broad multi-section/multi-fragment dispersion; reject paratext-only and quotation-dominated cases in the non-cued branch.
RescueIf no quotation or paratext cue and destination fanout is zero, rescue short near misses with span 450–870 and either coverage/section-gap or sum-coverage/section-count conditions.
Table 2. Final rule cascade used by the retained workflow. Spans are in characters; fanout is the number of distinct source sections strongly linked to the same target document.
 MainHardAll-labeled
MethodPRF1PRF1PRF1
Naive heuristic0.8510.9500.8980.1250.2730.1710.6720.8110.735
Structural-only0.9820.9170.9480.2310.1360.1710.8900.7300.802
Context-aware0.7780.9330.8480.3120.2270.2630.7520.7660.759
Decision tree0.8331.0000.9090.0410.1360.0630.5680.8290.674
LLM text-only0.3411.0000.5080.2170.9090.3510.3570.9820.524
Table 3. Main results on labeled ECCO–ECCO slices. All methods are scored against the same gold labels: 60, 22, and 111 positives in main , hard , and all-labeled .
Method#PosARMRHR
Naive heuristic9610.8110.9500.273
Structural-only7100.7300.9170.136
Context-aware1,2650.7660.9330.227
Decision tree1,2420.8291.0000.136
LLM text-only14,8860.9821.0000.909
LLM structured11,2230.9371.0000.773
Table 4. Deployment on the full ECCO–ECCO universe. #Pos = predicted positive count, restricted to the 22,735 shared pair universe; AR/MR/HR = recall on the all-labeled , main , and hard anchor sets.
ItemValue
Predicted positives (fully annotated)176, all genuine
Predicted negatives (audited)49
Resolved as non-reprint38
Resolved as reprint8
Uncertain3
P / R / F1 on resolved audit cases1.000 / 0.957 / 0.978
Table 5. Manual audit of ECCO–Newspaper results.
More in this topic

Rule-Based Classification

See all 21 papers

Explore similar work

CardsList
Aug 19, 2026

Institutional Newspapers Pipeline: Deriving billions of high quality tokens from historical newspapers

Historical newspapers are an abundant record of public life, but their dense, irregular and sometimes noisy layouts make computational access to these materials both challenging and limited. We present the Institutional Newspapers Pipeline, a modular system we jointly designed with Boston Public Library to extract high-quality, structured datasets from historical newspaper scans. It was architected so that each step remains interpretable and customizable, and so that the pipeline as a whole remains computationally frugal enough to run on workstation-level hardware. The pipeline runs each scan through a multi-step process: it segments scans into individual type-agnostic crops and performs OCR on each resulting segment before then performing text analysis, type classification, reading order detection, named entities recognition, subject classification, language detection, and pre-computed embeddings generation on every crop. We ran this pipeline against a portion of Boston Public Library's holdings and released the results as an open dataset. The optical character recognition (OCR) output represents 16.3 billion o200k_base tokens across 83.1 million individual crops, extracted from 1,473,635 public domain newspaper scans published between 1795 and 1930. This report describes our methods for each processing step, the small models we trained, as well as the evaluation results and dataset-scale measurements we collected in the process. It accompanies the release of the pipeline, models, and dataset. We position this work as a substantial step towards unlocking high-quality data from tens of millions of newspaper scans.

Matteo Cargnelutti, Catherine Brobston, Eben English +6

Institutional Data Initiative, Harvard Law School Library · Boston Public Library · Harvard Law School Library +1

Historical Document ProcessingDocument Information ExtractionDocument Layout Analysis
AskPDF
May 10, 2026

Matching Meaning at Scale: Evaluating Semantic Search for 18th-Century Intellectual History through the Case of Locke

While digitized corpora have transformed the study of intellectual transmission, current methods rely heavily on lexical text reuse detection, capturing verbatim quotations but fundamentally missing paraphrases and complex implicit engagement. This paper evaluates semantic search in 18th-century intellectual history through the reception of John Locke's foundational work. Using expert annotation grounded in a semantic taxonomy, we examine whether an off-the-shelf semantic search pipeline can surface meaning-level correspondences overlooked by lexical methods. Our results demonstrate that semantic search retrieves substantially more implicit receptions than lexical baselines. However, linguistic diagnostics also reveal a "lexical gatekeeping" effect, where retrieval remains partially constrained by surface vocabulary overlap. These findings highlight both the potential and the limitations of semantic retrieval for analyzing the circulation of ideas in large historical corpora. The data is available at https://github.com/COMHIS/locke-sim-data.

Yu Wu, Ananth Mahadevan, Filip Ginter +2

University of Helsinki · 2TurkuNLP, University of Turku · 3ELLIS Institute Finland

Historical Document AnalysisSemantic Textual SimilaritySemantic SearchIR Evaluation
AskPDF
May 25, 2026

Double Triangle Annotation: A Scalable Human-in-the-Loop Framework for High-Precision Historical Document Annotation

Evaluating structured-information extraction from historical documents at scale requires high-precision ground-truth annotations, yet traditional manual labeling is expensive and fully automated pipelines built on large language models are prone to hallucination. We propose Double Triangle Annotation, a two-layer human-in-the-loop framework that leverages cross-model consensus to automate the majority of annotation work while ensuring high-precision outputs. In the first layer, two architecturally independent Multimodal Large Language Models annotate each document in parallel; when they agree, the label is auto-accepted, and disagreements are routed to a human jury. A second layer cross-checks two such systems against each other, escalating residual conflicts to a domain expert. The framework rests on a single assumption -- error independence between models -- requires no distributional priors or task-specific calibration, and becomes more autonomous as model capability improves. On the Guides Rosenwald, a corpus of French medical directories spanning 1887-1906, the framework achieves a final Word Error Rate of 0.003. Applied at scale, model consensus auto-accepts over 85% of 13,595 fields. We release the resulting benchmark -- the first structured-extraction ground truth for the Rosenwald Guides -- to support future work on historical document processing.

Yi Ren

École Polytechnique Fédérale de Lausanne

Historical Document AnalysisLLM-Assisted AnnotationHuman-in-the-Loop AnnotationDocument Information ExtractionLanguage Model EnsemblesHistorical Document Processing
AskPDF