In the landscape of movie reviews and ratings, traditional tagging systems on platforms like IMDb, Rotten Tomatoes, and Letterboxd primarily cater to the explicit content of movies, often overlooking the rich undercurrents of audience perspectives and their thematic nuances. This project seeks to unearth these hidden voices by analyzing audience reviews, particularly focusing on five movies directed by Asian women—a segment frequently marginalized in mainstream discourse. The project challenges traditional digital environments on movie search, recommendation, and review by presenting voices from audiences in an alternative space.
Utilizing topic modeling, this project aims to look into the reviews of ten movies on four platforms (IMDb, Rotten Tomatoes, Letterboxd, Amazon) to identify emerging themes and patterns in reviews. By doing so, the project aspires to create new pathways for engaging with cinema through reviews, especially for lesser-known movies that might not receive widespread attention.
The choice of the topic modeling module Scikit-learn over alternatives like Gensim is due to my familiarity and the results from my experiments. I also deliberately exclude any rating information and refrain from conducting sentiment analysis to predict movie ratings. The results generated using Gensim are too detailed and descriptive considering the purpose of this project. As I continue to learn and test various topic modeling methods, I welcome feedback and insights that can refine and enhance this process.
Expected Outcomes:
Enhanced Accessibility and Discoverability: By spotlighting thematic elements within audience reviews, this project intends to elevate the visibility of movies directed by Asian women directors, making them more accessible to a broader audience. When viewers click on tags to explore reviews, they are presented with reviews from both popular and lesser-known movies, all categorized under the same thematic topic.
Increased Engagement: Several years ago, IMDb closed its forums. This project aims to provide a thematic exploration of film reviews, offering viewers a novel approach to understanding and appreciating cinema. It aims to foster deeper engagement with both the content and context of these films, as well as connect people who share similar interests.
A Digital Resource for Film Enthusiasts and Researchers: By documenting the methodology and sharing the code, this project will serve as a digital resource for film enthusiasts and researchers interested in Asian diaspora film studies. The open-source nature of the project encourages replication and adaptation, promoting a collaborative approach to film study.
The current website serves as a technical demonstration, showcasing a proof of concept using JSON files compiled by me. In the next stage, I plan to utilize MySQL and test machine learning classifiers to handle larger datasets and enhance data presentation. A basic classifier has already been designed and can be reviewed in the accompanying notebook. My demo represents a collective, non-hierarchical mode of sharing opinions and resources against the norms established by mainstream movie platforms. This project not only highlights the power of audience, fans, and reviews, but also emphasizes the importance of an inclusive digital space where diverse cinematic voices are recognized and celebrated.
Return to Tags