双11自然语言处理推荐

from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import linear_kernel

# 假设有一个商品列表及其描述
products = [
    {"id": 1, "name": "Laptop", "description": "High performance laptop with long battery life."},
    {"id": 2, "name": "Smartphone", "description": "Latest smartphone with advanced camera features."},
    # ... 其他商品
]

# 提取描述用于向量化
descriptions = [p["description"] for p in products]

# 使用TF-IDF向量化描述
tfidf = TfidfVectorizer(stop_words='english')
tfidf_matrix = tfidf.fit_transform(descriptions)

# 计算余弦相似度矩阵
cosine_sim = linear_kernel(tfidf_matrix, tfidf_matrix)

def get_recommendations(product_id, cosine_sim=cosine_sim):
    idx = next((i for i, p in enumerate(products) if p["id"] == product_id), None)
    if idx is None:
        return []
    
    sim_scores = list(enumerate(cosine_sim[idx]))
    sim_scores = sorted(sim_scores, key=lambda x: x[1], reverse=True)
    sim_scores = sim_scores[1:6]  # 获取最相似的前5个商品索引
    product_indices = [i[0] for i in sim_scores]
    return products[product_indices]

# 示例调用
recommended_products = get_recommendations(1)  # 假设用户对商品ID为1的商品感兴趣
print(recommended_products)