向量化（Embedding）

import dashscope
import numpy as np
from dashscope import TextEmbedding

# 如果使用華北2（北京）地區的模型，需要將base_url替換為：https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1
dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'

def cosine_similarity(a, b):
    """計算餘弦相似性"""
    return np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))

def semantic_search(query, documents, top_k=5):
    """語義搜尋"""
    # 產生查詢向量
    query_resp = TextEmbedding.call(
        model="qwen3.7-text-embedding",
        input=query,
        dimension=1024
    )
    query_embedding = query_resp.output['embeddings'][0]['embedding']

    # 產生文檔向量
    doc_resp = TextEmbedding.call(
        model="qwen3.7-text-embedding",
        input=documents,
        dimension=1024
    )

    # 計算相似性
    similarities = []
    for i, doc_emb in enumerate(doc_resp.output['embeddings']):
        similarity = cosine_similarity(query_embedding, doc_emb['embedding'])
        similarities.append((i, similarity))

    # 排序並返回top_k結果
    similarities.sort(key=lambda x: x[1], reverse=True)
    return [(documents[i], sim) for i, sim in similarities[:top_k]]

# 使用樣本
documents = [
    "人工智慧是電腦科學的一個分支",
    "機器學習是實現人工智慧的重要方法",
    "深度學習是機器學習的一個子領域"
]
query = "什麼是AI？"
results = semantic_search(query, documents, top_k=2)
for doc, sim in results:
    print(f"相似性: {sim:.3f}, 文檔: {doc}")

import dashscope
import numpy as np
from dashscope import TextEmbedding

# 如果使用華北2（北京）地區的模型，需要將base_url替換為：https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1
dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'

def cosine_similarity(a, b):
    """計算餘弦相似性"""
    return np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))
def build_recommendation_system(user_history, all_items, top_k=10):
    """構建推薦系統"""
    # 產生使用者歷史向量
    history_resp = TextEmbedding.call(
        model="qwen3.7-text-embedding",
        input=user_history,
        dimension=1024
    )

    # 計算使用者偏好向量（取平均）
    user_embedding = np.mean([
        emb['embedding'] for emb in history_resp.output['embeddings']
    ], axis=0)

    # 產生所有物品向量
    items_resp = TextEmbedding.call(
        model="qwen3.7-text-embedding",
        input=all_items,
        dimension=1024
    )

    # 計算推薦分數
    recommendations = []
    for i, item_emb in enumerate(items_resp.output['embeddings']):
        score = cosine_similarity(user_embedding, item_emb['embedding'])
        recommendations.append((all_items[i], score))

    # 排序並返回推薦結果
    recommendations.sort(key=lambda x: x[1], reverse=True)
    return recommendations[:top_k]

# 使用樣本
user_history = ["科幻類", "動作類", "懸疑類"]
all_movies = ["未來世界", "太空探險", "古代戰爭", "浪漫之旅", "超級英雄"]
recommendations = build_recommendation_system(user_history, all_movies)
for movie, score in recommendations:
    print(f"推薦分數: {score:.3f}, 電影: {movie}")

# 需要安裝 scikit-learn: pip install scikit-learn
import dashscope
import numpy as np
from sklearn.cluster import KMeans

# 如果使用華北2（北京）地區的模型，需要將base_url替換為：https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1
dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'

def cluster_texts(texts, n_clusters=2):
    """將一組文本進行聚類"""
    # 1. 擷取所有文本的向量
    resp = dashscope.TextEmbedding.call(
        model="qwen3.7-text-embedding",
        input=texts,
        dimension=1024
    )
    embeddings = np.array([item['embedding'] for item in resp.output['embeddings']])

    # 2. 使用KMeans演算法進行聚類
    kmeans = KMeans(n_clusters=n_clusters, random_state=0, n_init='auto').fit(embeddings)

    # 3. 整理並返回結果
    clusters = {i: [] for i in range(n_clusters)}
    for i, label in enumerate(kmeans.labels_):
        clusters[label].append(texts[i])
    return clusters

# 使用樣本
documents_to_cluster = [
    "手機公司A發售新款手機",
    "搜尋引擎公司B推出新款系統",
    "世界盃決賽阿根廷對陣法國",
    "奧運會中國隊再添一金",
    "某公司發布最新AI晶片",
    "歐洲杯賽事報道"
]
clusters = cluster_texts(documents_to_cluster, n_clusters=2)
for cluster_id, docs in clusters.items():
    print(f"--- 類別 {cluster_id} ---")
    for doc in docs:
        print(f"- {doc}")

import dashscope
import numpy as np

# 如果使用華北2（北京）地區的模型，需要將base_url替換為：https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1
dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'

def cosine_similarity(a, b):
    """計算餘弦相似性"""
    return np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))

def classify_text_zero_shot(text, labels):
    """零樣本文本分類"""
    # 1. 擷取輸入文本和所有標籤的向量
    resp = dashscope.TextEmbedding.call(
        model="qwen3.7-text-embedding",
        input=[text] + labels,
        dimension=1024
    )
    embeddings = resp.output['embeddings']
    text_embedding = embeddings[0]['embedding']
    label_embeddings = [emb['embedding'] for emb in embeddings[1:]]

    # 2. 計算與每個標籤的相似性
    scores = [cosine_similarity(text_embedding, label_emb) for label_emb in label_embeddings]

    # 3. 返回相似性最高的標籤
    best_match_index = np.argmax(scores)
    return labels[best_match_index], scores[best_match_index]

# 使用樣本
text_to_classify = "這件衣服的料子很舒服，款式也好看"
possible_labels = ["數位產品", "服裝配飾", "食品飲料", "家居生活"]

label, score = classify_text_zero_shot(text_to_classify, possible_labels)
print(f"輸入文本: '{text_to_classify}'")
print(f"最匹配的分類是: '{label}' (相似性: {score:.3f})")

import dashscope
import numpy as np

# 以下為新加坡地區的配置，調用時請將WorkspaceId替換為真實的業務空間ID，各地區的配置不同。
dashscope.base_http_api_url = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1"

def cosine_similarity(a, b):
    """計算餘弦相似性"""
    return np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))

def detect_anomaly(new_comment, normal_comments, threshold=0.6):
    # 1. 向量化所有正常評論和新評論
    all_texts = normal_comments + [new_comment]
    resp = dashscope.TextEmbedding.call(
        model="qwen3.7-text-embedding",
        input=all_texts,
        dimension=1024
    )
    embeddings = [item['embedding'] for item in resp.output['embeddings']]

    # 2. 計算正常評論的中心向量（平均值）
    normal_embeddings = np.array(embeddings[:-1])
    normal_center_vector = np.mean(normal_embeddings, axis=0)

    # 3. 計算新評論與中心向量的相似性
    new_comment_embedding = np.array(embeddings[-1])
    similarity = cosine_similarity(new_comment_embedding, normal_center_vector)

    # 4. 判斷是否為異常
    is_anomaly = similarity < threshold
    return is_anomaly, similarity

# 使用樣本
normal_user_comments = [
    "今天的會議很有成效",
    "專案進展順利",
    "下周發布新版本",
    "使用者反饋良好"
]

test_comments = {
    "正常評論": "功能符合預期",
    "異常-無意義亂碼": "asdfghjkl zxcvbnm"
}

print("--- 異常檢測樣本 ---")
for desc, comment in test_comments.items():
    is_anomaly, score = detect_anomaly(comment, normal_user_comments)
    result = "是" if is_anomaly else "否"
    print(f"評論: '{comment}'")
    print(f"是否為異常: {result} (與正常樣本相似性: {score:.3f})\n")