An interpretable multi-transformer ensemble for text-based movie genre classification
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
Multi-label movie genre classification is challenging due to the inherent ambiguity and overlap between different genres. Most of the existing works in genre classification use audio-visual modalities. The potential of text-based modalities in movie genre classification is still underexplored. This paper proposes an ensemble deep-learning model that uses movie plots to predict movie genres. After pre-processing the text plots, three transformer-based models, Bidirectional Encoder Representations from Transformers (BERT), DistilBERT, and Robustly Optimized BERT Pre-training Approach (ROBERTa), are used to generate genre predictions, combined through a weighted soft-voting method. The proposed ensemble architecture achieves state-of-the-art performance on two benchmark datasets, Trailers12K and LMTD9, with a micro-average precision of 80.10% and 80.37%, respectively, significantly outperforming both traditional machine learning approaches and advanced deep learning models. The ensemble's superior performance is attributed to its ability to combine the diverse strengths of individual models and capture nuanced genre-specific information from textual features. The lack of interpretability in deep learning models for genre classification is addressed using Local Interpretable Model-Agnostic Explanations (LIME), which provides both local and global explanations for the model's predictions. The findings of the study highlight the potential of textual data in automated genre classification and emphasize the importance of interpretability methods in multi-label genre classification.
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 | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| 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 it