A Text Extraction Software Benchmark Based on a Synthesized Dataset
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
Text extraction plays an important function for data processing workflows in digital libraries. For example, it is a crucial prerequisite for evaluating the quality of migrated textual documents. Complex file formats make the extraction process error-prone and have made it very challenging to verify the correctness of extraction components. Based on digital preservation and information retrieval scenarios, three quality requirements in terms of effectiveness of text extraction tools are identified: 1) is a certain text snippet correctly extracted from a document, 2) does the extracted text appear in the right order relatively to other elements and, 3) is the structure of the text preserved. A number of text extraction tools is available fulfilling these three quality requirements to various degrees. However, systematic benchmarks to evaluate those tools are still missing, mainly due to the lack of datasets with accompanying ground truth. The contribution of this paper is two-fold. First we describe a dataset generation method based on model driven engineering principles and use it to synthesize a dataset and its ground truth directly from a model. Second, we define a benchmark for text extraction tools and complete an experiment to calculate performance measures for several tools that cover the three quality requirements. The results demonstrate the benefits of the approach in terms of scalability and effectiveness in generating ground truth for content and structure of text elements.
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.
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 | 0.000 | 0.003 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| 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