Labeling Consistency Test of Multi-Label Data for Aspect and Sentiment Classification Using the Cohen Kappa Method
Bibliographic record
Abstract
Classification is part of machine learning, and developing it requires labeled data. Most data is available in an unlabeled form. Data labeling is a step that researchers must take. Good labeled data will produce a good classification model. The data labeling process cannot be ignored and needs to be done carefully and consistently. Because the classification process requires well-labeled data that can be accounted for. In addition, good labeled data will produce a good classification model. The role of an expert (rater) is needed to label the data and ideally at least two experts. However, involving two raters will become a new problem because it is likely that the results of the inter-rater labeling will be different. We propose the Cohen Kappa method to overcome this problem. We used data from scraping user reviews of the Indonesian marketplace, there were 4.307. Based on the calculation results, Kappa=0.909 for aspect detection, Kappa=0.893 for sentiment classification, and Kappa=0.971 for class aspect. Based on the kappa value, the labeling results for aspect detection, sentiment classification and aspect class were declared "almost perfect agreement", so that the results of this research obtained labeled data that can be used for classification tasks, especially for developing aspect-based sentiment analysis models.
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 | 0.108 | 0.298 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.003 |
| Bibliometrics | 0.008 | 0.005 |
| Science and technology studies | 0.005 | 0.003 |
| Scholarly communication | 0.004 | 0.004 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
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".