GraphLink

This Web-Application was created to demonstrate and analyse the graph theory based recommender model (https://github.com/05kashyap/SocialNet_RecSys).

It has been hosted at: http://kashyap05.pythonanywhere.com/

Note: This application was created as a part of our discrete mathematics course project.

Project Report:

Feature Overview:

-User registration and authentication.

-Reset Password

-Inbuilt graph recommender system with 3 cases. (Details can be found in graph/alterutil.py)

-Social Network graph updated in real time with community detection.

-Follow users

-Create Posts

-Like and commend under posts

-personalized "your feed" to view content from users who are being followed

-Featured section with top liked posts

ScreenShots

Control flow of the recommendations system

Real-Time graph generated using a matplotlib + networkx backend

Recommended users carousel

Results

Evaluation Metrics	Result
Accuracy	0.7877
Precision (at k=5)	1.0
Recall (at k=5)	0.538095
F1 Score	0.6996

Graph Matching Metrics	Result
GED	0.6
Edge overlap ratio	0.91044
Recall (at k=5)	0.538095
Structural Hammering Distance	0.19999

Tech Stack

Django, Python, Matplotlib, Networkx, SQlite3

Setup Instructions(To run the application locally): (Ideally inside a virtual env)

• Clone the repository

  git clone [https://github.com/05kashyap/GDSC_meme_feed](https://github.com/05kashyap/GraphLink.git)

• Install requirements.txt

  pip install -r requirements.txt

• Make migrations

  python3 manage.py makemigrations

• Apply migrations

  python3 manage.py migrate

• Apply migrations

  python3 manage.py runserver