Classifying domain-specific text documents containing ambiguous keywords
Bibliographic record
Abstract
A keyword-based search of comprehensive databases such as PubMed may return irrelevant papers, especially if the keywords are used in multiple fields of study. In such cases, domain experts (curators) need to verify the results and remove the irrelevant articles. Automating this filtering process will save time, but it has to be done well enough to ensure few relevant papers are rejected and few irrelevant papers are accepted. A good solution would be fast, work with the limited amount of data freely available (full paper body may be missing), handle ambiguous keywords and be as domain-neutral as possible. In this paper, we evaluate a number of classification algorithms for identifying a domain-specific set of papers about echinoderm species and show that the resulting tool satisfies most of the abovementioned requirements. Echinoderms consist of a number of very different organisms, including brittle stars, sea stars (starfish), sea urchins and sea cucumbers. While their taxonomic identifiers are specific, the common names are used in many other contexts, creating ambiguity and making a keyword search prone to error. We try classifiers using Linear, Naïve Bayes, Nearest Neighbor, Tree, SVM, Bagging, AdaBoost and Neural Network learning models and compare their performance. We show how effective the resulting classifiers are in filtering irrelevant articles returned from PubMed. The methodology used is more dependent on the good selection of training data and is a practical solution that can be applied to other fields of study facing similar challenges. Database URL: The code and date reported in this paper are freely available at http://xenbaseturbofrog.org/pub/Text-Topic-Classifier/.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.009 | 0.040 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.036 | 0.029 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.005 | 0.003 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.007 | 0.004 |
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 source (direct Gemma or distilled Codex), 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".