Recommender systems plays an important role in many web services. Thanks to them Internet platforms can generate maximum profit and their clients be as satisfied as possible. In my thesis I focus on comparison of algorithms and then I implement the best one in form of a web application. Neural networks are used as well due to their increasing popularity in recommendation tasks.
On index page app user has 4 main choices of recommendations:
- collaborative filtering where was used baseline model
- neural networks model with one hidden layer
- knn based algorithm
- content based algorithm
When user send request to a server by filling a form he or she is redirected to the new page with the best recommendations.
Best tab shows level of similarity between user and item.
All recomendations are shown to user based on his or her location (if the user hasn't changed the original one).
Application is fully responsive. Due to that smart choice can be accesed from every device.
user_id: YqMpcRUA0OMw1WNLDGoj-A, business_id: p0iEUamJVp_QpaheE-Nz_g service name: Roaring Fork city: Las Vegas categories: choose your own