MOOD

Participants

• Joanna Blaszka
• Kamil Dutkiewicz
• Paulina Jewulska (team leader)
• Dawid Justyna
• Riza Yarar

Do you need more people: No

Short description of the idea

MOOD is a simple web application that checks users' mood and choose adequate entertainment depending on their choices. Funny movie, good book or interesting podcast? Users only need to answer some questions about their feelings and time limitation and app quickly selects a lot of options that hopely satisfy user.

How to start

1. Create virtual environment: python -m venv mood-env
2. Activate virtual environment:
   • on Linux: source mood-env/bin/activate
   • on Windows: .\mood-env\Scripts\activate
3. Install dependencies: pip install -r requirements.txt
4. Run application: python manage.py runserver

Sentiment analisys:

• for user mood analysis we are using tool vaderSentiment but for getting sentiment score for movies we are using word2vect model we decided to proceed in thi way becouse the scrpit word2vect.py is too slow to give back sentiment and don’t have sense wait 3 / 4 minute for get user mood

usefull link:

• for word2vect model : https://radimrehurek.com/gensim/models/word2vec.html
• for dataset aclImdb :http://ai.stanford.edu/~amaas/data/sentiment/
• guide used for understanding how to implement word2vect : https://towardsdatascience.com/machine-learning-word-embedding-sentiment-classification-using-keras-b83c28087456

Datasets:

• tmdb_movies_data.csv: used for creating tmdb_movies.json

• movie_data.csv used for traint and test model word2vect ,this file contain 50k reviews of films with sentiment score 0(negative) or 1(positive) so we thought Is suitable for our case for creating this dataset from "aclImdb dataset" we can use this scripts makecsv.py but we are not sure to use this scripts becouse movie_data is 65.9 mb so it will be more fast if we put csv file in project folder

• for correct working of word2vect.py script we must run also scripts/downloadmodule.py that install additional module to our work enviorement

Sentiment analisys:

• for user mood analysis we are using tool vaderSentiment but for getting sentiment score for movies we are using word2vect model we decided to proceed in thi way becouse the scrpit word2vect.py is too slow to give back sentiment and don’t have sense wait 3 / 4 minute for get user mood

usefull link:

• for word2vect model : https://radimrehurek.com/gensim/models/word2vec.html
• for dataset aclImdb :http://ai.stanford.edu/~amaas/data/sentiment/
• guide used for understanding how to implement word2vect : https://towardsdatascience.com/machine-learning-word-embedding-sentiment-classification-using-keras-b83c28087456

Datasets:

• tmdb_movies_data.csv: used for creating tmdb_movies.json

• movie_data.csv used for traint and test model word2vect ,this file contain 50k reviews of films with sentiment score 0(negative) or 1(positive) so we thought Is suitable for our case for creating this dataset from "aclImdb dataset" we can use this scripts makecsv.py but we are not sure to use this scripts becouse movie_data is 65.9 mb so it will be more fast if we put csv file in project folder

• for correct working of word2vect.py script we must run also scripts/downloadmodule.py that install additional module to our work enviorement

Technology stack:

• Frontend: Django Templates, Bootstrap
• Backend: Python, Django
• Database: SQLite

Illustrations by https://icons8.com.

Features

• User registration

Whenever user want, can go back to previous results.

• Adding a calendar to mark the mood

MOOD will adapt the form of entertainment to the user for each day.

• Adding preferences

The user can help the application in the selection of appropriate entertainment.

• Matching movies with 'Netflix' and 'HBO'

Using ML, we can match movies to the user, so that he can see what he wants

• Matching podcasts, books, movies and series

Register user can talk to the bot and match the result more closely.

• Statistics about preferences of users using

time range, age range, Sex, Period of year(winter,spring,etc...)

User Stories:

• as a user I would like to be able to log in with my personal account
• as a user I would like to enter the password twice during registration so as not to make mistakes
• as a user I would like to see the rating of the film I got
• as a user I would like to see a list of movies in the database
• as a user I would like to rate movies

Project roadmap:

SPRINT 1:

• Create trello board
• Create github repository

SPRINT 2:

• Learn Django and set up project
• Create basic layout
• Study datasets
• Learn word2vec

SPRINT 3:

• Brainstorm on questions
• Add new implementations to layout
• Collect data from user (not logged)
• Preprocessing of dataset
• Write some unit tests

SPRINT 4:

• Classification of dataset
• Get emotions from https://www.kaggle.com/lakshmi25npathi/imdb-dataset-of-50k-movie-reviews
• Get similar themes from that dataset
• Use word2vec to find similar emotions and themes in that sample
• Divide movies into emotions categories using previously mention sample
• Divide movies into themes categories

SPRINT 5:

• Implement of matching function
• Display results to the user (not logged)
• Add registration and login
• Write some tests

SPRINT 6:

• Creation of statistics and displaying them
• Write some tests

SPRINT 7:

• Final corections and test

Added by mwmajew:

• average movie score for imdb and average score for the user input