Cross-lingual text alignment for fine-grained plagiarism detection
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Fast and easy access to a wide range of documents in various languages, in conjunction with the wide availability of translation and editing tools, has led to the need to develop effective tools for detecting cross-lingual plagiarism. Given a suspicious document, cross-lingual plagiarism detection comprises two main subtasks: retrieving documents that are candidate sources for that document and analysing those candidates one by one to determine their similarity to the suspicious document. In this article, we examine the second subtask, also called the detailed analysis subtask, where the goal is to align plagiarised fragments from source and suspicious documents in different languages. Our proposed approach has two main steps: the first step tries to find candidate plagiarised fragments and focuses on high recall, followed by a more precise similarity analysis based on dynamic text alignment that will filter the results by finding alignments between the identified fragments. With these two steps, the proximity of the terms will be considered in different levels of granularity. In both steps, our approach uses a dictionary to obtain translations of individual terms instead of using a machine translation system to convert longer passages from one language to another. We used a weighting scheme to distinct multiple translations of the terms. Experimental results show that our method outperforms the methods used by the systems that achieved the best results in the PAN-2012 and PAN-2014 competitions.
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.
Direct model labels (unvalidated)
Per-model category and study-design labels from the labeling rounds. They are machine output, unvalidated, and the disagreement between models ships as data. No study design here is MEDLINE-validated yet.
| Model arm | Categories | Study design | Confidence |
|---|---|---|---|
| gemma | Research integrity Domain: not available · Genre: Empirical About the Canadian research system: no · About a Canadian topic: no | Simulation or modeling | high |
| gpt | no category Domain: not available · Genre: Methods About the Canadian research system: no · About a Canadian topic: no | Other design | low |
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 | 0.003 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.008 |
| Open science | 0.001 | 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 it