Roadmap: Recommendation Engine

At time of writing, posts (and comments) on the home page of the website are ranked either by recency or by number of upvotes. How can we use posts that a user has previously liked to recommend them posts more relevant to their needs?

Roadmap

1. Goal

1.1. Asks users for interests when the log in for the first time (they can choose to skip)
1.2. Learns from the users actions (upvotes, bookmarks, downvotes, clicks, searches, etc)
1.3. Uses that information to rerank the users feed and search results to be more personalized
1.4. Last step (Optional): Upgrades to ML-based recommendations when enough data can/is collected (clicks, searches, etc)

2. Feature Roadmap

2.1. Interest Onboarding (must be completed in order a-e)

• a. Pop-up that shows up when the user first logs in
• b. Create relevant tags/interests that the user can click
  • b1. User has the option to select multiple fields of CS/CompE/related fields
• c. Create "Skip" button --> doesn't save selected interests
  • c1. Default value for interests is []
• d. Create "Continue" button --> saves selected interests
  • d1. Stores interests in user data
• e. Edit interests later (Settings/Profile)

2.2. Implicit Interest Learning

• Track and log: upvote, bookmark, downvote, post view, search query + clicked result
• Update per-user interest profile based on tagged content they engage with

2.3. Better Recommendations in Search Results

• Rerank search results using the same interest profile scoring

2.4. Periodical Email Digest of Relevant Posts and Events

2.5. Optional: ML-Based Recommendations (Later)