Movie Recommendation System

If you want to run the code, you need to make sure that your environment statisfies the following requirement:

• Python 3
• Django
• pymsql, numpy, pandas, sklearn, tensorflow
• MySQL

Our project is tested on Ubuntu 18.04 x64. If your system environment is different with us, it may cause some problems.

Download

You need to download our pre-processed dataset ml-latest-small from 360Yun ( ml-latest-small.zip ) which contains our pre-processed and itermediate data structure. Put the folder ml-latest-small in Movierecommend\users\rating\.

We write three kinds of ranking algorithms:

• naive_ranking
• ranking
• dfm_ranking

Look up the funtion ranking(data, mode, df) in view.py. You can send the parameter into the funtion to test different methods.

mode = 'naive' # naive ranking model
     = 'rank'  # ranking model
     = 'dfm'   #  dfm model

df   = 'user'  # user-based collaborative filter
     = 'item'  # item-based collaborative filter 
     = 'both'  # intersection of two collaborative filters

To run the second model, it need to pre train and the trained model can be downloaded at 360Yun ( save.zip ). If you want to test it, please decompress it to users\rating.

Configure database

Create a database and then configure the SQL connection settings in django_auth_example\settings.py. Modify the get_conn() function in views.py.

At the root of the project, run

python manage.py migrate 

to migrate the database tables. Moreover, you have to create a new table moviegenre and import the ml-latest-small\process\moviegenre.csv into it.

Run server

Finally, after doing all of these, you can run the server by

python manage.py runserver