The app has been implemented using the Gradio python library. The app execution file is named as deployment.ipynb. Once the code in this file is run, the app interface can be interacted either in the jupyter notebook or in the browser using the local host address that is given by Gradio after running the code.
• datasets.ipynb – The jupyter notebook used to analyse and pre-process the three datasets used in this project. • training process.ipynb – Jupyter notebook used to run three iterations of training process. • scrapper.ipynb – Jupyter notebook used to scrape the screenplays from the IMSdb website. • classification.ipynb – Jupyter notebook used to classify the scenes into emotions and store them as a vector. • deployment.ipynb – Jupyter notebook used to deploy the web interface app. • classifier – the folder were the trained model and tokenizer configurations are stored. • crowdflower dataset – folder where the crowdflower dataset csv file is stored. • GoEmotions – Folder where the GoEmotions csv dataset file is stored. • dair-ai emotion dataset – Folder where the DAIR AI emotion dataset is stored. • final dataset – Folder where the pre-processed datasets are stored. • screenplay datasets – Folder where the scrapped screenplays csv and json file and the emotion classified scene dataset is stored. • harry potter and the sorcerer’s stone.txt – File used to test the analyse screenplay function.
the classifier file and the screenplay data file was more than 100 mb so it couldn't be uploaded hence a google drive link to download the file has been given to upload it. classifier model file link - https://drive.google.com/drive/folders/1FtilXQgv2zrHMsk2HE8uD-h9CjrSScoI?usp=drive_link
screenplay datasets -