System Architecture

System Architecture Overview

This architecture outlines the flow of token data and user interactions within the JujuConnect platform. It highlights the relationship between services that manage user data, video streams, recommendation services, and how tokens are utilized throughout these processes.

Services & Data Flow

The architecture consists of several key components and services:

1. User Interaction (Flutter App):

   • The user's actions (like watching videos) are central to the token distribution process.
   • The user interacts with various services like Upload Service and Recommendation Service that handle video data and personalized recommendations.

2. Transfer Service (R2 Stream Redirect):

   • This service handles adaptive bitrate streaming, video encoding, and live streaming support.
   • It also provides optimization for video delivery, ensuring users experience smooth playback.

3. Upload Service:

   • This service manages the upload of videos. It works closely with the Extract Service to process videos after they are uploaded.

4. Extract Service & Extraction Job:

   • The Extract Service analyzes uploaded videos and extracts metadata.
   • The extraction job runs a process to create embeddings and vectors for the uploaded videos.

5. Recommendation Service:

   • Based on user interactions and watch history, this service processes recommendations for the user.
   • It uses cached data for new users (recommendation cache) and logged-in users to deliver personalized content.

6. Databases:

   • Videos DB: Stores all the video metadata.
   • Watch History DB: Tracks user interactions and preferences.
   • Neighbour DB: Keeps track of video relationships for recommendation purposes.
   • Extract DB: Stores the extracted video metadata.
   • Vector DB: Stores video embeddings for recommendations and other personalized content.

Tokenomics Flow

1. User Activity: When a user interacts with the platform, their watch history and engagement data are logged. This data influences recommendations and token distribution, ensuring content creators are rewarded based on engagement.

2. Recommendation System: As part of JujuConnect's dynamic recommendation system, users are shown personalized video content. The better the content performs, the more tokens are distributed to the creators.

3. Token Distribution: Tokens are earned based on user interaction, including:

   • Video Plays: Tokens are rewarded for each view based on the current rate.
   • Engagement & Interaction: Tokens can also be awarded based on user actions like likes, comments, and shares, ensuring creators are compensated for content that engages the community.