About

Welcome to SRec, a website for AI/ML-based video game recommendations available on the Steam store. While it‘s easy to find recommendations for popular games, it could be tricky for less popular games. SRec aims to provide decent recommendations for all kinds of games. SRec also provides review insight of Steam games based on user reviews.

Although Steam already offers recommendations through its Interactive Recommender and More Like This (click for an example) feature, sometimes it‘s just not good enough. To tackle the limitation, SRec offers two recommendation systems that provide different varieties of similar games:

• A tag-based recommendation system, which shows explanation by showing most similar tags.
• A smart recommendation system utilizing Steam user preferences, game tags, and other metadata.

SRec is developed by Sandy Khosasi. You can see my other works on GitHub and HuggingFace. All data shown on this website are obtained from Steam. I use PyTorch to train recommender models and TorchServe to serve trained models. Additionally, I utilize PostgreSQL to manage hundreds of millions of rows of data and meet the requirements for creating some SRec features.