Sentiment based movie recommender system

High level architecture

Components

1) LSTM-based sentiment recommender chatbot trained on Twitter data

• Dataset used - Figure 8 labelled textual dataset

• Model summary :

• List of emotions detected :

Index	Emotion
0	anger
1	boredom
2	empty
3	enthusiasm
4	fun
5	happiness
6	hate
7	love
8	neutral
9	relief
10	sadness
11	surprise
12	worry

• Model Evaluation

• Model Accuracy : 37%

2) Cosine Similarity based Movie Recommendation System

• Dataset used : The Movies Dataset
• Vectorizer used : Count Vectorizer
• Cosine Similarity used to recommend movie IDs based on movie title

Frameworks used

The frontend has been developed in React JS & the backend consists of APIs made in Django Rest Framework & Python. The project has been dockerised and near deployment.

Preview

https://cosmic-empanada-cfc5e2.netlify.app/ (Frontend only)