Citation segmentation from sparse & noisy data: A joint inference approach with Markov logic networks
Bibliographic record
Abstract
This article presents an approach to citation segmentation that addresses special challenges as typically found in Digital Humanities applications. We perform citation segmentation from Optical Character Recognition (OCR) input obtained from volumes of a printed bibliography, the Turkology Annual . This showcase application features serious difficulties for state-of-the-art techniques in citation segmentation: multilingual citation entries , lack of data redundancy , inconsistencies , and noise from OCR input . Our approach is based on Markov logic networks (MLN) (Richardson and Domingos, Markov logic networks. Machine Learning , 62 (1): 107–36, 2006), a framework of statistical relational learning that combines first-order logic with probabilistic modeling. Formalization in first-order logic offers high expressivity and flexibility, and makes it possible to tailor segmentation to specific conventions of a given bibliography. We show that in face of the specific difficulties found with segmenting references from a digitized bibliography, our MLN formalizations outperform state-of-the-art statistical methods. We obtain 88% F 1 -score for exact field match, a 24.8% increase over a conditional random fields-based system baseline. In contrast to prior work, we address a data set featuring sparse and noisy data. Our method extends Poon and Domingos (Joint Inference in information extraction. In Proceedings of the Twenty-Second National Conference on Artificial Intelligence . Vancouver, Canada: AAAI Press, 2007)’s approach by applying joint inference at the field level . By this move, we are able to cope with the lack of citation redundancy and noise in the data. Our approach can be characterized as knowledge-based and hence does not rely on annotated training data. The rule sets we designed can be adapted to other bibliographies, or further types of digitized sources, such as historical dictionaries or encyclopedias.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.003 | 0.004 |
| Open science | 0.002 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".