Database

All data safely on MongoDB Atlas.
All read/write operations must have the correct authentication token.

Collection Schemas

1. Users

We use the users collection to track all details about a user.
However, we do not authenticate users through our database, as that would be insecure. 👍

users_model.ts

export interface UserDocument extends Document {
  firstName: string
  lastName: string
  username: string
  email: string
  password: string
  birthDate: Date
  registrationDate: Date
  lastLoginDate: Date
  profilePicture: string
  onBoardingStatus: string
  emailVerificationCode: number
  isAdmin: boolean
  comparePassword(candidatePassword: string): Promise<boolean>
}

2. Clips

The clips collection tracks clips metadata, allowing users to see and engage with video proxies before starting to stream the video.
It also provides the streaming endpoint for the video.

clip_model.ts

export interface ClipMetadataDocument extends Document {
  videoId: mongoose.Types.ObjectId
  title: string
  description: string
  uploadDate: Date
  uploader: mongoose.Types.ObjectId
  duration: number
  thumbnailURL: string
  clipURL: string
  views: mongoose.Types.ObjectId[]
  likes: mongoose.Types.ObjectId[]
  dislikes: mongoose.Types.ObjectId[]
}

3. Comments

The comments collection tracks comments and replies on videos, allowing users to share their sentiments about videos.

comment_model.ts

export interface CommentDocument extends Document {
  userId: mongoose.Types.ObjectId
  videoId: mongoose.Types.ObjectId
  clipId: mongoose.Types.ObjectId | null
  text: string
  likes: mongoose.Types.ObjectId[]
  nestedComments: mongoose.Types.ObjectId[]
  isReply: boolean
}

The recommendations collection tracks which videos are in the recommended queue for a given user, allowing the front-end to prefetch metadata and even prerender the next video while the user is still watching the current video.
This reduces buffering, providing a smoother user experience.

recommendation_model.ts

export interface PrecomputedRecommendationsDocument extends Document {
  userId: mongoose.Types.ObjectId
  topicSequences: Map<string, mongoose.Types.ObjectId[]>
}

5. Relationships

The relationships collection tracks relationshipts between users, allowing users to connect with others and share content.
Annoying users can also be blocked.

relationship_model.ts

export interface RelationshipDocument extends Document {
  fromUserID: mongoose.Types.ObjectId
  toUserID: mongoose.Types.ObjectId
  status: 'pending' | 'accepted' | 'declined' | 'blocked'
  initiatedDate: Date
  updatedDate: Date
}

6. Topics

The topics collection tracks all topics and subtopics in our ecosystem, allowing us to organize videos effectively, and efficiently direct users through learning paths.

topic_model.ts

export interface TopicMetadataDocument extends Document {
  combinedTopicName: string
  topicName: string
  subTopicName: string | undefined
  description: string
  thumbnailURL: string
}

7. User Affinities

The user-affinities collection tracks user affinities to topics, depending on how much they have engaged with videos in that topic, and the sentiments reflected in their feedback.

user_affinity_model.ts

export interface UserAffinityDocument extends Document {
  userId: mongoose.Types.ObjectId
  affinities: Map<mongoose.Types.ObjectId, number>
}

8. Video Affinities

The video-affinities collection tracks video affinities to topics, depending on how much each video aligns with the topic.

This is because certain videos might be aligned with multiple topics.

video_affinity_model.ts

export interface VideoAffinityDocument extends Document {
  videoId: mongoose.Types.ObjectId
  affinities: Map<mongoose.Types.ObjectId, number>
}

9. Video Metadata

The video-metadata collection tracks all metadata individual long-form videos and which clips were extracted from each specific video.

We recognize that clips from the same video might be related.

video_model.ts

export interface VideoMetadataDocument extends Document {
  title: string
  description: string
  uploadDate: Date
  uploader: mongoose.Types.ObjectId
  duration: number
  thumbnailURL: string
  topicId: mongoose.Types.ObjectId
  clips: mongoose.Types.ObjectId[]
  views: mongoose.Types.ObjectId[]
  likes: mongoose.Types.ObjectId[]
  dislikes: mongoose.Types.ObjectId[]
}

Back-End

The Discite Backend Engine.

Video Streaming

Nuxt provides several component layers to implement the user interface of your application.

Data Model