We have built a recommendation system using user's foe and friends liked products.
Most of the social network utilize the friend's information we utilized both friends and foe information. Our motivation for this product is to utilize negative links(foe) for the recommendation system.
There can be two types of information in the graph to be used in the recommendation system:
- Local Information: recommendation using the friend's and foes' likes and dislikes
- Global Information: recommendation using an influenced celebrity likes and dislike
We have used both the information and created the optimization function and derive a derivative to minimize the error and cost.
Results are shown in the picture. We have found that user choices are closer to the friends and farther to the foes. With our implementation, we have also proved it.
Uses: The recommendation system can be used to give better recommendation to the user in terms of products, movies, music or videos, it not only considers the positive but also the negative link.
Log in or sign up for Devpost to join the conversation.