Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Predicting Oscar-Nominated Screenplays with Sentence Embeddings

Francis Gross

TL;DR

This work investigates predicting Oscar nominations for screenplays using modern long-text embeddings. By constructing the Movie-O-Label dataset from MovieSum scripts and Oscar records, and evaluating a simple pipeline with E5-based embeddings and logistic regression, it demonstrates that combining script, summary, and title signals yields non-trivial predictive performance (ROC-AUC ~0.79, Macro-F1 ~0.66). The study highlights that text-derived signals can inform nomination likelihood, while acknowledging limitations such as data size, language scope, and non-text factors. It also outlines practical, reproducible steps and points to future work involving larger models, multimodal data, and broader award coverage to enhance predictive capability and industry relevance.

Abstract

Oscar nominations are an important factor in the movie industry because they can boost both the visibility and the commercial success. This work explores whether it is possible to predict Oscar nominations for screenplays using modern language models. Since no suitable dataset was available, a new one called Movie-O-Label was created by combining the MovieSum collection of movie scripts with curated Oscar records. Each screenplay was represented by its title, Wikipedia summary, and full script. Long scripts were split into overlapping text chunks and encoded with the E5 sentence em bedding model. Then, the screenplay embed dings were classified using a logistic regression model. The best results were achieved when three feature inputs related to screenplays (script, summary, and title) were combined. The best-performing model reached a macro F1 score of 0.66, a precision recall AP of 0.445 with baseline 0.19 and a ROC-AUC of 0.79. The results suggest that even simple models based on modern text embeddings demonstrate good prediction performance and might be a starting point for future research.

Predicting Oscar-Nominated Screenplays with Sentence Embeddings

TL;DR

This work investigates predicting Oscar nominations for screenplays using modern long-text embeddings. By constructing the Movie-O-Label dataset from MovieSum scripts and Oscar records, and evaluating a simple pipeline with E5-based embeddings and logistic regression, it demonstrates that combining script, summary, and title signals yields non-trivial predictive performance (ROC-AUC ~0.79, Macro-F1 ~0.66). The study highlights that text-derived signals can inform nomination likelihood, while acknowledging limitations such as data size, language scope, and non-text factors. It also outlines practical, reproducible steps and points to future work involving larger models, multimodal data, and broader award coverage to enhance predictive capability and industry relevance.

Abstract

Oscar nominations are an important factor in the movie industry because they can boost both the visibility and the commercial success. This work explores whether it is possible to predict Oscar nominations for screenplays using modern language models. Since no suitable dataset was available, a new one called Movie-O-Label was created by combining the MovieSum collection of movie scripts with curated Oscar records. Each screenplay was represented by its title, Wikipedia summary, and full script. Long scripts were split into overlapping text chunks and encoded with the E5 sentence em bedding model. Then, the screenplay embed dings were classified using a logistic regression model. The best results were achieved when three feature inputs related to screenplays (script, summary, and title) were combined. The best-performing model reached a macro F1 score of 0.66, a precision recall AP of 0.445 with baseline 0.19 and a ROC-AUC of 0.79. The results suggest that even simple models based on modern text embeddings demonstrate good prediction performance and might be a starting point for future research.

Paper Structure

This paper contains 14 sections, 4 equations, 3 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methods
  4. Solving the Issue.
  5. Data
  6. Solution Approach and Methodology
  7. Chunking.
  8. Embeddings.
  9. Classification.
  10. Model Variants.
  11. Implementation and Reproducibility
  12. Results
  13. Discussion
  14. Conclusion and Future Work

Figures (3)

  • Figure 1: Compact task pipeline.
  • Figure 2: ROC-AUC of the best-performing model, Script+Summary+Title.
  • Figure 3: PR-AUC of the best-performing model, Script+Summary+Title.