💡 Ideas for the Algorithm

[Vianpyro]

Hey team! 👋

As we continue developing TindETS, it’s time to brainstorm ideas for our matchmaking algorithm! The approach we choose will play a crucial role in user engagement and satisfaction, so let’s get creative and deliberate. Below are some questions and ideas to guide our discussion:

---

🤔 Questions to consider:

1. User Preferences:

   • Should we allow users to set filters (e.g., age, department, hobbies) or let the algorithm handle everything?
   • How should we prioritize these preferences?
   • Should users be able to rank preferences (e.g., hobbies > faculty)?

2. Engagement Factors:

   • Should we prioritize active users or give everyone equal visibility?
   • How do we ensure less popular profiles also get matches?

3. Matching Logic:

   • What factors should be prioritized in matching (e.g., mutual preferences, similar interests, engagement history)?
   • Should the algorithm include an element of randomness to encourage discovery?

4. Diversity in Matches:

   • Should we introduce randomness to avoid monotony and expand user discovery?

5. Feedback Loop:

   • Should we adapt matches based on user interactions (e.g., likes, views, chats)?

6. Fairness & Popularity:

   • How do we avoid over-favoring highly "liked" users?
   • How do we ensure less popular users still get matches while keeping the app engaging for everyone?

7. Personalization:

   • How important is it to show users why they matched (e.g., "You both love hiking!")?

8. Data Privacy:

   • How do we balance personalized matching with user data privacy and security?

9. Simplicity vs. Complexity:

   • Do we want a basic algorithm to start with, or should we aim for something more sophisticated like machine learning?

---

💡 Ideas for the algorithm:

Here are some potential elements for our matchmaking algorithm:

• Scoring System:
  Rank matches based on factors like mutual interests, shared preferences, proximity, and engagement history.

• Dynamic Prioritization:
  Boost visibility for users who have fewer matches to keep the platform fair.

• Recency Bias:
  Favor recently active users to encourage responsiveness and improve real-time interactions.

• Serendipity Factor:
  Introduce a small amount of randomness to diversify matches and increase engagement.

---

🚀 Technical Considerations:

We can implement the algorithm as part of our backend, using SQL queries for match scoring and Python for more complex logic. Here’s a basic example:

-- Example matchmaking query
SELECT user1.id AS user_id, user2.id AS match_id,
       (10 * mutual_interests_count +
        5 * CASE WHEN user1.faculty = user2.faculty THEN 1 ELSE 0 END +
        3 * proximity_score) AS match_score
FROM users AS user1
JOIN users AS user2
  ON user1.id != user2.id
WHERE user1.active = TRUE
  AND user2.active = TRUE
ORDER BY match_score DESC;

---

🗳️ Let’s talk!

What do you think? Share your ideas for how the algorithm should work or vote for any of the ideas mentioned above. If you have other suggestions or concerns about scalability, fairness, or engagement, let us know!

Let’s work together to build the perfect matchmaking experience for TindETS!