类型安全的推荐系统：内容发现类型实现的深入探讨

            
type Movie = {
  id: string;
  title: string;
  genres: string[];
  releaseYear: number;
  director: { name: string; nationality: string };
  ratings: {
    imdb: number;
    rottentomatoes: number;
  };
};

type Article = {
  id: string;
  headline: string;
  author: { name: string };
  publicationDate: Date;
  tags: string[];
  url: string;
};

// A union type to represent any content item
type ContentItem = Movie | Article;

function processContentItem(item: ContentItem): void {
  if (item.hasOwnProperty('releaseYear')) { // Type guard to narrow down to Movie
    const movie = item as Movie; // Or use a more robust type guard
    console.log(`Processing movie: ${movie.title} released in ${movie.releaseYear}`);
    // Access movie-specific properties safely
    movie.genres.forEach(genre => console.log(`- Genre: ${genre}`));
  } else if (item.hasOwnProperty('headline')) { // Type guard for Article
    const article = item as Article;
    console.log(`Processing article: ${article.headline} published on ${article.publicationDate}`);
    // Access article-specific properties safely
    article.tags.forEach(tag => console.log(`- Tag: ${tag}`));
  }
}

            

              
                Copy
              
            
          

            
from pydantic import BaseModel
from typing import List, Optional
from datetime import datetime

class Director(BaseModel):
    name: str
    nationality: str

class Movie(BaseModel):
    id: str
    title: str
    genres: List[str]
    release_year: int
    director: Director
    ratings: dict  # Can be further refined with nested models

class Article(BaseModel):
    id: str
    headline: str
    author_name: str
    publication_date: datetime
    tags: List[str]
    url: str

# Example of data validation
raw_movie_data = {
    "id": "m123",
    "title": "Inception",
    "genres": ["Sci-Fi", "Action"],
    "release_year": 2010,
    "director": {"name": "Christopher Nolan", "nationality": "British"},
    "ratings": {"imdb": 8.8, "rottentomatoes": 0.87}
}

try:
    movie_instance = Movie(**raw_movie_data)
    print(f"Successfully validated movie: {movie_instance.title}")
except Exception as e:
    print(f"Data validation failed: {e}")

# Example of invalid data
invalid_movie_data = {
    "id": "m456",
    "title": "The Matrix",
    "genres": "Sci-Fi", # Incorrect type, should be a list
    "release_year": 1999,
    "director": {"name": "Lana Wachowski", "nationality": "American"},
    "ratings": {"imdb": 8.7, "rottentomatoes": 0.88}
}

try:
    movie_instance = Movie(**invalid_movie_data)
except Exception as e:
    print(f"Data validation failed for invalid data: {e}") # This will catch the error

            

              
                Copy
              
            
          

            
import numpy as np
from typing import Dict, List, Tuple, Optional

# Define type for embeddings
Embedding = np.ndarray

class RecommendationModel:
    def __init__(self, embedding_dim: int):
        self.embedding_dim = embedding_dim
        self.user_embeddings: Dict[str, Embedding] = {}
        self.item_embeddings: Dict[str, Embedding] = {}

    def get_user_embedding(self, user_id: str) -> Optional[Embedding]:
        return self.user_embeddings.get(user_id)

    def get_item_embedding(self, item_id: str) -> Optional[Embedding]:
        return self.item_embeddings.get(item_id)

    def generate_recommendations(self, user_id: str, top_n: int = 10) -> List[str]:
        user_emb = self.get_user_embedding(user_id)
        if user_emb is None:
            return []

        # Calculate similarity scores (e.g., cosine similarity)
        scores: List[Tuple[str, float]] = []
        for item_id, item_emb in self.item_embeddings.items():
            # Ensure embeddings have the correct shape and type for calculation
            if user_emb.shape[0] != self.embedding_dim or item_emb.shape[0] != self.embedding_dim:
                print(f"Warning: Mismatched embedding dimension for {item_id}")
                continue
            if user_emb.dtype != np.float32 or item_emb.dtype != np.float32: # Example type check
                print(f"Warning: Unexpected embedding dtype for {item_id}")
                continue
            
            similarity = np.dot(user_emb, item_emb) / (np.linalg.norm(user_emb) * np.linalg.norm(item_emb))
            scores.append((item_id, similarity))

        # Sort and get top N items
        scores.sort(key=lambda x: x[1], reverse=True)
        recommended_item_ids = [item_id for item_id, score in scores[:top_n]]
        return recommended_item_ids

# Example usage (assuming embeddings are pre-loaded/trained)
# model = RecommendationModel(embedding_dim=64)
# model.user_embeddings['user1'] = np.random.rand(64).astype(np.float32)
# model.item_embeddings['itemA'] = np.random.rand(64).astype(np.float32)
# recommendations = model.generate_recommendations('user1')

            

              
                Copy
              
            
          

            
from enum import Enum
from pydantic import BaseModel
from typing import Optional
from datetime import datetime

class InteractionType(Enum):
    VIEW = 1
    CLICK = 2
    LIKE = 3
    RATING = 4
    PURCHASE = 5

class InteractionRecord(BaseModel):
    user_id: str
    item_id: str
    interaction_type: InteractionType
    timestamp: datetime
    value: Optional[float] = None # For RATING or other quantifiable interactions

def process_interaction(record: InteractionRecord):
    if record.interaction_type == InteractionType.RATING:
        if record.value is None or not (0 <= record.value <= 5): # Example: check value range
            print(f"Warning: Invalid rating value for user {record.user_id}, item {record.item_id}")
            return
        # Process rating
        print(f"User {record.user_id} rated item {record.item_id} with {record.value}")
    elif record.interaction_type in [InteractionType.VIEW, InteractionType.CLICK, InteractionType.LIKE, InteractionType.PURCHASE]:
        # Process binary interactions
        print(f"User {record.user_id} performed {record.interaction_type.name} on item {record.item_id}")
    else:
        print(f"Unknown interaction type: {record.interaction_type}")

# Example usage
rating_interaction = InteractionRecord(
    user_id="userA",
    item_id="itemB",
    interaction_type=InteractionType.RATING,
    timestamp=datetime.now(),
    value=4.5
)
process_interaction(rating_interaction)

view_interaction = InteractionRecord(
    user_id="userA",
    item_id="itemC",
    interaction_type=InteractionType.VIEW,
    timestamp=datetime.now()
)
process_interaction(view_interaction)

            

              
                Copy