Introduction

The LikeMinds Feed is a dynamic content delivery system designed to ensure users receive the most relevant and engaging posts. It operates through three different types of feeds, each catering to specific user needs and content delivery strategies. These are:

1. Universal Feed – A simple, chronologically sorted feed.
2. Ranked Feed – A feed ranked based on user engagement and recency.
3. Personalized Feed – A highly customized feed based on user preferences and behaviors.
4. Topic Feed - A feed aggregated based on the topic selected/followed by the user.

This document provides a detailed overview of each type of feed, their sorting mechanisms, and how they contribute to the overall LikeMinds experience.

1. Universal Feed

Overview

The Universal Feed is the simplest feed type, where posts are displayed in the order they were created. This feed ensures that users see the latest content first without any ranking modifications.

Sorting Algorithm

• Posts are arranged in reverse chronological order (newest to oldest).
• No influence of engagement metrics like likes or comments.
• Ideal for real-time updates where recency is the key factor.

Use Cases

• Communities that prefer a raw and unfiltered timeline.
• Users who want to see the latest content without algorithmic ranking.

2. Ranked Feed

Overview

The Ranked Feed enhances user experience by prioritizing posts based on engagement and freshness. This approach ensures that the most interacted-with content gains higher visibility.

Sorting Algorithm

The feed ranks posts using a scoring system based on three key metrics:

1. Likes – Higher liked posts receive a better ranking.
2. Comments – Posts with more comments are considered more engaging.
3. Recency – Newer posts are given a ranking boost to maintain freshness.

The ranking formula can be represented as:

Score =(Likes/MaxLikes)* LikesWeight+(Comments/MaxComments)* CommentsWeight+(Recency/MaxRecency)*RecencyWeight

Where:

Likes

• Likes → Number of likes on the post
• MaxLikes → The highest possible number of likes considered for ranking.
  • Beyond this limit, the Likes count contributes the highest possible score to the ranking score.
• LikesWeight → The weight assigned to likes
  • This can be customised based on Customer requirement

Comments

• Comments → Number of comments on the post
• MaxComments → The highest possible number of comments considered for ranking.
  • Beyond this limit, the Comments count contributes the highest possible score to the ranking score.
• CommentsWeight → The weight assigned to comments
  • This can be customised based on Customer requirement

Recency

• Recency → How recently the post was created (in minutes)
• MaxRecency → The highest possible recency value considered for ranking.
  • Beyond this limit, the Recency count contributes the highest possible score to the ranking score.
• RecencyWeight → The weight assigned to recency
  • The recency factor follows a time decay function, ensuring that older posts gradually lose rank over time.
  • This can be customised based on Customer requirement

Use Cases

• Feed Applications that want to highlight trending discussions.
• Users who prefer seeing popular posts rather than purely the latest ones.

3. Personalized Feed

Overview

The Personalized Feed takes customization to the next level by tailoring content based on user preferences and interactions. It builds upon the Ranked Feed by incorporating user-specific behaviors.

Sorting Algorithm

The personalization algorithm enhances ranking by adding three more dimensions:

1. Topics Followed – Posts from topics a user follows receive a ranking boost.
2. Users Followed – Posts from users a person follows are prioritized.
3. Engagement Metrics – The standard ranked feed metrics (Likes, Comments, Recency) are still considered.

The ranking formula expands to:

Score = Ranking feed Score + (Topics Followed * Topics Weight)+(Users Boost * Users Weight) + Post Seen

Where:

Ranking Feed Score

The score is calculated based on the combination of Likes, Comments, and Recency as mentioned above.

Dampening

• Post seen → The post has been viewed by the user in the current session and will be removed from the personalised feed.
  • Post dampening will be removed after 24 hours.

Topics Boost

• Topics Followed → The topics a user follows
• TopicsWeight → The weight assigned to topic relevance in ranking
  • This can be customised based on the Customer requirement

Users Boost

• Users Followed → The users a person follows
• UsersWeight → The weight assigned to followed users in ranking
  • This can be customised based on the Customer requirement

Use Cases

• Feed Application where users have diverse interests.
• Personalized content delivery for higher engagement.
• Users who want a balance between trending and relevant content.

Conclusion

The LikeMinds Feed system provides flexibility in content discovery through its three-tiered approach. While the Universal Feed ensures real-time updates, the Ranked Feed enhances visibility of popular content, and the Personalized Feed ensures highly relevant posts reach the user. The combination of these approaches allows LikeMinds to cater to a wide range of community needs, improving engagement and user satisfaction.