Skip to main content

Crate oximedia_recommend

Crate oximedia_recommend 

Source
Expand description

Content recommendation and discovery engine for OxiMedia.

oximedia-recommend provides comprehensive recommendation capabilities for media platforms, including content-based filtering, collaborative filtering, hybrid approaches, and advanced personalization features.

§Features

  • Content-based Filtering: Recommend similar content based on features
  • Collaborative Filtering: User behavior-based recommendations
  • Hybrid Approach: Combine multiple recommendation methods
  • Similarity Metrics: Calculate content similarity using various metrics
  • User Profiles: Build and manage user preference profiles
  • View History: Track and analyze viewing patterns
  • Rating System: Handle explicit and implicit ratings
  • Trending Detection: Identify trending content in real-time
  • Personalization: Context-aware personalized recommendations
  • Diversity: Ensure recommendation diversity and avoid filter bubbles
  • Freshness: Balance popular and new content

§Modules

§Example

use oximedia_recommend::{RecommendationEngine, RecommendationRequest};
use uuid::Uuid;

// Create a recommendation engine
let engine = RecommendationEngine::new();

// Get recommendations for a user
let user_id = Uuid::new_v4();
let request = RecommendationRequest {
    user_id,
    limit: 10,
    ..Default::default()
};

// let recommendations = engine.recommend(&request)?;

Re-exports§

pub use error::RecommendError;
pub use error::RecommendResult;

Modules§

ab_test
A/B testing framework for recommendation algorithms.
als
Alternating Least Squares (ALS) Matrix Factorization for collaborative filtering.
bandits
Multi-armed bandit algorithms for exploration/exploitation in recommendations.
batch_recommend
Batch recommendation generation for offline/pre-computation scenarios.
calibration
Recommendation score calibration.
cold_start
Cold start strategies for recommendation systems.
collab_filter
Collaborative filtering utilities: user-item matrix operations and similarity metrics.
collaborative
Collaborative filtering algorithms.
content
Content-based filtering and similarity calculation.
content_based
Content-based recommendation helpers.
content_filter
Content-based filtering utilities.
context_signal
Contextual signals for context-aware media recommendations.
contextual_bandits
Contextual bandit algorithms for exploration/exploitation in live recommendations.
cross_domain
Cross-domain recommendation engine.
decay_model
Time-decay models for aging user preferences and interactions.
dense_linalg
Pure Rust dense linear algebra types replacing ndarray dependency.
diversity
Diversity enforcement.
diversity_rerank
Diversity-aware re-ranking via Maximal Marginal Relevance (MMR).
embargo
Content embargo and scheduling for time-gated recommendations.
error
Error types for recommendation engine.
evaluation
Offline evaluation metrics for recommendation quality.
explain
Recommendation explanation generation.
exploration_policy
Exploration vs exploitation policies for the recommendation system.
fairness
Recommendation fairness metrics and exposure equity enforcement.
feature_store
Feature storage and retrieval for recommendation models.
federated
Federated learning support for collaborative recommendation models.
feedback_signal
Feedback signal processing for recommendation systems.
freshness
Fresh content promotion.
genre_affinity
Genre affinity modeling with temporal decay.
history
View history tracking and analysis.
hybrid
Hybrid recommendation approaches.
impression_tracker
Track content impressions and compute click-through / engagement rates.
item_similarity
Item vector similarity primitives for content-based recommendation.
knowledge_graph
Knowledge graph-based recommendations.
lsh
Locality-Sensitive Hashing (LSH) for approximate nearest-neighbour search.
multi_objective
Multi-objective optimization for recommendation ranking.
novelty
Novelty and familiarity scoring for content recommendations.
personalize
Personalization engine.
playlist_generator
Automatic playlist generation from a seed item.
popularity_bias
Popularity bias detection and correction for recommendation systems.
profile
User profile management.
rank
Ranking and scoring.
ranking
Ranking algorithms for recommendation lists.
rate_limit
Token-bucket rate limiter for the recommendation engine.
rating
Rating system (explicit and implicit).
recommendation_score
Weighted scoring primitives for composing recommendation relevance signals.
score_cache
Score caching layer for recommendation systems.
sequence_model
Sequential recommendation modeling.
session
Session-based recommendation engine.
session_recommend
Session-based recommendation engine.
svd_pp
SVD++ — Enhanced Matrix Factorization with Implicit Feedback.
trending
Trending content detection.
trending_detection
Trending content detection via exponentially-weighted view velocity.
user_profile
User profile management for recommendation personalization.
user_segment
User segment-based recommendations.
watch_history
Watch history tracking and genre-preference analysis.

Structs§

ContentMetadata
Content metadata for recommendations
DiversitySettings
Diversity settings for recommendations
Recommendation
Recommendation result item
RecommendationContext
Context information for recommendations
RecommendationEngine
Main recommendation engine coordinating all recommendation capabilities
RecommendationRequest
Recommendation request configuration
RecommendationResults
Recommendation results

Enums§

RecommendationReason
Reason for recommendation
RecommendationStrategy
Recommendation strategy
TimeOfDay
Time of day categories