Module learning 

Self-learning engine using GNN-inspired approaches.

This module implements adaptive learning mechanisms inspired by Graph Neural Networks to improve search results over time based on user interactions.

Components

• LearningEngine: Core learning with GNN message passing
• FeedbackCollector: Implicit signal capture from user interactions
• BatchScheduler: Background jobs for pattern detection and enrichment
• InsightStore: Storage for discovered patterns and relationships

Learning Flow

User Interactions
       │
       ▼
┌─────────────────┐
│FeedbackCollector│ ──► Implicit signals (view, select, dwell)
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│FeedbackProcessor│ ──► ProcessedFeedback (relevance deltas)
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│ LearningEngine  │ ──► Update weights, GNN propagation
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│ BatchScheduler  │ ──► Patterns, gaps, classifications
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  InsightStore   │ ──► Published insights for retrieval
└─────────────────┘

Re-exports

pub use batch::BatchInput;

pub use batch::BatchJob;

pub use batch::BatchScheduler;

pub use batch::EntryMetadata;

pub use batch::Insight;

pub use batch::InsightStore;

pub use batch::JobRun;

pub use batch::JobStatus;

pub use batch::JobType;

pub use batch::KnowledgeClass;

pub use batch::RelationshipType;

pub use batch::Trend;

pub use feedback::FeedbackCollector;

pub use feedback::FeedbackConfig;

pub use feedback::FeedbackProcessor;

pub use feedback::FeedbackSignal;

pub use feedback::ProcessedFeedback;

pub use feedback::QueryId;

pub use feedback::SessionId;

pub use feedback::SignalType;

Modules

batch

Batch Learning Jobs

feedback

Implicit Feedback Collection

Structs

GnnLayer

A simplified GNN layer for embedding transformation.

LearningEngine

Learning engine that improves search quality over time.

LearningStats

Learning statistics.

ReplayBuffer

Experience replay buffer using reservoir sampling.