Shopee 的 item_review 接口是获取平台商品评论数据的核心接口，能够返回特定商品的用户评价、评分、图文内容等关键信息。对于跨境电商卖家而言，这些数据对于了解用户反馈、优化产品描述、改进产品质量以及制定营销策略具有重要价值。
一、接口核心特性分析
- 接口功能与定位
-
核心功能：获取指定商品的用户评论数据，支持分页和筛选
-
数据特点：
- 包含多语言评论内容（ Shopee 支持的主要语言）
- 带有用户评分、购买属性、评论时间等元数据
- 包含用户上传的评论图片
- 支持筛选有图评论、特定评分区间评论
- 包含评论的点赞数和回复数等互动数据
-
应用场景：
- 产品质量反馈分析
- 用户满意度评估
- 竞品评论对比分析
- 评论内容情感分析
- 产品改进方向挖掘
- 认证机制
Shopee 开放平台的认证机制相对复杂，主要包括：
- 基于
partner_id和shopid的店铺身份验证 - 通过
access_token进行 API 访问授权 - 请求需要包含签名（
sign）参数，确保请求合法性 - 部分市场的评论数据需要额外申请权限
- 核心参数与响应结构
请求参数
| 参数名 | 类型 | 是否必填 | 说明 |
|---|---|---|---|
item_id |
Long | 是 | 商品 ID |
page |
Integer | 否 | 页码，默认 1 |
page_size |
Integer | 否 | 每页条数，默认 20，最大 50 |
rating_filter |
Integer | 否 | 评分筛选：0 (全部)、1-5 (特定评分) |
with_images |
Boolean | 否 | 是否只看有图评论，默认 false |
language |
String | 否 | 评论语言筛选，如 zh-CN, en, th 等 |
partner_id |
Integer | 是 | 合作伙伴 ID |
shopid |
Integer | 是 | 店铺 ID |
access_token |
String | 是 | 访问令牌 |
timestamp |
Long | 是 | 时间戳 (毫秒) |
sign |
String | 是 | 请求签名 |
响应核心字段
-
分页信息：总评论数、总页数、当前页码
-
评论列表：每条评论包含
- 基础信息：评论 ID、用户 ID、用户昵称
- 评分信息：商品评分、物流评分、服务评分
- 内容信息：评论文字、评论时间、评论语言
- 多媒体：评论图片 URL 列表
- 商品信息：购买的变体属性、购买时间
- 互动数据：点赞数、回复数、卖家回复内容
二、Python 脚本实现
以下是调用 Shopee item_review 接口的完整 Python 实现，包含签名生成、请求处理、数据解析和深度分析功能： import requests import time import json import logging import hashlib import hmac import os import re from typing import Dict, Optional, List, Tuple from requests.exceptions import RequestException import matplotlib.pyplot as plt import numpy as np import pandas as pd from collections import defaultdict from textblob import TextBlob # 用于情感分析
配置日志 logging.basicConfig( level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s" )
class ShopeeItemReviewAPI: def init(self, partner_id: int, partner_key: str, shopid: int, region: str = "sg"): """ 初始化Shopee评论接口客户端 :param partner_id: 合作伙伴ID :param partner_key: 合作伙伴密钥 :param shopid: 店铺ID :param region: 地区代码，如sg(新加坡), my(马来西亚), th(泰国), tw(台湾) """ self.partner_id = partner_id self.partner_key = partner_key.encode('utf-8') # 用于签名的密钥 self.shopid = shopid self.region = region self.base_url = self._get_base_url() self.access_token = None self.session = requests.Session() self.session.headers.update({ "Content-Type": "application/json", "Accept": "application/json", "User-Agent": "ShopeeAPI/1.0.0 (Python)" })
python
def _get_base_url(self) -> str:
"""根据地区获取基础URL"""
region_map = {
"sg": "https://partner.shopeemobile.com", # 新加坡
"my": "https://partner.shopeemobile.com", # 马来西亚
"th": "https://partner.shopeemobile.com", # 泰国
"tw": "https://partner.shopee.tw", # 台湾
"id": "https://partner.shopeemobile.com", # 印度尼西亚
"vn": "https://partner.shopeemobile.com", # 越南
"ph": "https://partner.shopeemobile.com" # 菲律宾
}
return region_map.get(self.region, region_map["sg"])
def _generate_signature(self, path: str, timestamp: int, access_token: str = "") -> str:
"""生成请求签名"""
# 构建待签名字符串
base_string = f"{self.partner_id}{path}{timestamp}"
if access_token:
base_string += access_token
# 使用HMAC-SHA256计算签名
signature = hmac.new(
self.partner_key,
base_string.encode('utf-8'),
hashlib.sha256
).hexdigest()
return signature
def set_access_token(self, access_token: str) -> None:
"""设置访问令牌"""
self.access_token = access_token
def get_item_reviews(self,
item_id: int,
page: int = 1,
page_size: int = 20,
rating_filter: int = 0,
with_images: bool = False,
language: Optional[str] = None) -> Optional[Dict]:
"""
获取商品评论
:param item_id: 商品ID
:param page: 页码
:param page_size: 每页条数
:param rating_filter: 评分筛选：0(全部)、1-5(特定评分)
:param with_images: 是否只看有图评论
:param language: 评论语言筛选
:return: 评论数据
"""
# 验证参数
if page_size < 1 or page_size > 50:
logging.error(f"每页条数必须在1-50之间，当前为: {page_size}")
return None
if rating_filter < 0 or rating_filter > 5:
logging.error(f"评分筛选值必须在0-5之间，当前为: {rating_filter}")
return None
# 检查access_token
if not self.access_token:
logging.error("请先设置access_token")
return None
# 构建请求路径和参数
path = "/api/v2/item/get_reviews"
timestamp = int(time.time() * 1000) # 毫秒时间戳
signature = self._generate_signature(path, timestamp, self.access_token)
params = {
"partner_id": self.partner_id,
"shopid": self.shopid,
"item_id": item_id,
"page": page,
"page_size": page_size,
"rating_filter": rating_filter,
"with_images": with_images,
"timestamp": timestamp,
"access_token": self.access_token,
"sign": signature
}
# 添加可选参数
if language:
params["language"] = language
try:
url = f"{self.base_url}{path}"
response = self.session.get(url, params=params, timeout=15)
response.raise_for_status()
result = response.json()
# 检查响应状态
if result.get("error") != 0:
logging.error(f"获取评论失败: {result.get('message', '未知错误')} (错误码: {result.get('error')})")
return None
# 格式化评论数据
return self._format_review_data(result, item_id)
except RequestException as e:
logging.error(f"获取评论请求异常: {str(e)}")
return None
except json.JSONDecodeError:
logging.error(f"评论响应解析失败: {response.text[:200]}...")
return None
def _format_review_data(self, review_data: Dict, item_id: int) -> Dict:
"""格式化评论数据"""
# 分页信息
pagination = {
"total_reviews": review_data.get("data", {}).get("total_count", 0),
"total_pages": (review_data.get("data", {}).get("total_count", 0) +
review_data.get("data", {}).get("page_size", 20) - 1) //
review_data.get("data", {}).get("page_size", 20),
"current_page": review_data.get("data", {}).get("page", 1),
"page_size": review_data.get("data", {}).get("page_size", 20)
}
# 格式化评论列表
reviews = []
for review in review_data.get("data", {}).get("reviews", []):
# 处理评分
rating = review.get("rating", 0)
# 处理评论图片
images = []
if review.get("images"):
images = [img.get("url") for img in review.get("images") if img.get("url")]
# 处理变体信息
variant_info = ""
if review.get("variation"):
variant_info = ", ".join([v.get("name", "") for v in review.get("variation") if v.get("name")])
# 处理卖家回复
seller_reply = None
if review.get("seller_reply"):
seller_reply = {
"content": review["seller_reply"].get("comment", ""),
"created_time": review["seller_reply"].get("created_at", 0)
}
# 评论内容翻译（简单处理）
content = review.get("comment", "")
translated_content = self._translate_comment(content, review.get("language", ""))
# 情感分析
sentiment = self._analyze_sentiment(content or translated_content)
reviews.append({
"review_id": review.get("review_id"),
"item_id": item_id,
"user": {
"user_id": review.get("user_id"),
"username": review.get("author_username", "匿名用户")
},
"rating": rating,
"rating_star": "★" * rating + "☆" * (5 - rating),
"content": content,
"translated_content": translated_content,
"language": review.get("language", ""),
"sentiment": sentiment,
"images": images,
"variant_info": variant_info,
"purchase_time": review.get("purchase_time", 0),
"created_time": review.get("created_at", 0),
"like_count": review.get("like_count", 0),
"reply_count": review.get("cmt_count", 0),
"seller_reply": seller_reply,
"is_verified_purchase": review.get("is_verified", False)
})
return {
"pagination": pagination,
"reviews": reviews,
"raw_data": review_data # 保留原始数据
}
def _translate_comment(self, content: str, language: str) -> str:
"""简单翻译评论内容（实际应用中可使用专业翻译API）"""
if not content or language == "zh-CN":
return content
# 这里使用TextBlob进行简单翻译，实际应用中建议使用Google Translate API等专业服务
try:
if language in ["en", "zh-TW", "th", "id", "vi", "ms", "tl"]:
blob = TextBlob(content)
return str(blob.translate(to="zh-CN"))
return content
except:
# 翻译失败时返回原文
return content
def _analyze_sentiment(self, content: str) -> Dict:
"""分析评论情感倾向"""
if not content:
return {"polarity": 0, "subjectivity": 0, "sentiment": "neutral"}
# 使用TextBlob进行情感分析
blob = TextBlob(content)
polarity = blob.sentiment.polarity # 情感极性：-1(负面)到1(正面)
subjectivity = blob.sentiment.subjectivity # 主观性：0(客观)到1(主观)
# 确定情感类别
if polarity > 0.1:
sentiment = "positive"
elif polarity < -0.1:
sentiment = "negative"
else:
sentiment = "neutral"
return {
"polarity": round(polarity, 4),
"subjectivity": round(subjectivity, 4),
"sentiment": sentiment
}
def get_all_reviews(self, item_id: int, max_pages: int = 10, **kwargs) -> List[Dict]:
"""
获取商品的所有评论
:param item_id: 商品ID
:param max_pages: 最大页数限制
:return: 所有评论列表
"""
all_reviews = []
page = 1
while page <= max_pages:
logging.info(f"获取第 {page} 页评论")
result = self.get_item_reviews(
item_id=item_id,
page=page,
page_size=50, # 使用最大页大小
**kwargs
)
if not result or not result["reviews"]:
break
all_reviews.extend(result["reviews"])
# 检查是否已到最后一页
if page >= result["pagination"]["total_pages"]:
break
page += 1
# 控制请求频率，Shopee API通常限制QPS为10
time.sleep(0.1)
logging.info(f"共获取到 {len(all_reviews)} 条评论")
return all_reviews
def analyze_reviews(self, reviews: List[Dict]) -> Dict:
"""分析评论数据"""
if not reviews:
return {}
total = len(reviews)
# 评分分析
rating_analysis = self._analyze_ratings(reviews)
# 情感分析
sentiment_analysis = self._analyze_sentiments(reviews)
# 内容关键词分析
keyword_analysis = self._analyze_keywords(reviews)
# 变体分析
variant_analysis = self._analyze_variants(reviews)
# 图片分析
image_analysis = self._analyze_images(reviews)
# 卖家回复分析
reply_analysis = self._analyze_seller_replies(reviews)
# 提取典型评论
typical_reviews = self._get_typical_reviews(reviews)
return {
"total_reviews": total,
"rating_analysis": rating_analysis,
"sentiment_analysis": sentiment_analysis,
"keyword_analysis": keyword_analysis,
"variant_analysis": variant_analysis,
"image_analysis": image_analysis,
"reply_analysis": reply_analysis,
"typical_reviews": typical_reviews
}
def _analyze_ratings(self, reviews: List[Dict]) -> Dict:
"""分析评分分布"""
rating_counts = defaultdict(int)
total_rating = 0
for review in reviews:
rating = review["rating"]
rating_counts[rating] += 1
total_rating += rating
avg_rating = round(total_rating / len(reviews), 1) if reviews else 0
return {
"average": avg_rating,
"counts": dict(rating_counts),
"distribution": {k: round(v/len(reviews)*100, 1) for k, v in rating_counts.items()}
}
def _analyze_sentiments(self, reviews: List[Dict]) -> Dict:
"""分析情感分布"""
sentiment_counts = defaultdict(int)
polarity_scores = []
for review in reviews:
sentiment = review["sentiment"]["sentiment"]
sentiment_counts[sentiment] += 1
polarity_scores.append(review["sentiment"]["polarity"])
avg_polarity = round(sum(polarity_scores) / len(polarity_scores), 4) if polarity_scores else 0
# 按评分分组的情感分析
rating_sentiment = defaultdict(lambda: defaultdict(int))
for review in reviews:
rating_sentiment[review["rating"]][review["sentiment"]["sentiment"]] += 1
return {
"distribution": dict(sentiment_counts),
"ratio": {k: round(v/len(reviews)*100, 1) for k, v in sentiment_counts.items()},
"avg_polarity": avg_polarity,
"by_rating": {k: dict(v) for k, v in rating_sentiment.items()}
}
def _analyze_keywords(self, reviews: List[Dict]) -> Dict:
"""分析评论关键词"""
# 核心关键词类别
keywords = {
"质量": ["质量", "品质", "做工", "材料", "耐用", "差", "好"],
"物流": ["快递", "物流", "速度", "慢", "快", "包装", "运输"],
"尺寸": ["大小", "尺寸", "合适", "太大", "太小", "长短"],
"外观": ["颜色", "外观", "样子", "好看", "漂亮", "难看"],
"价格": ["价格", "划算", "便宜", "贵", "性价比"],
"功能": ["功能", "好用", "实用", "方便", "难用"]
}
keyword_counts = defaultdict(int)
negative_keywords = defaultdict(int)
for review in reviews:
# 使用翻译后的内容进行关键词分析
content = review["translated_content"] or review["content"]
if not content:
continue
content_lower = content.lower()
is_negative = review["sentiment"]["sentiment"] == "negative"
for category, kws in keywords.items():
for kw in kws:
if kw in content_lower:
keyword_counts[category] += 1
if is_negative:
negative_keywords[category] += 1
break # 每个类别只计数一次
# 计算负面关键词占比
keyword_negative_ratio = {}
for category, count in keyword_counts.items():
if count > 0:
keyword_negative_ratio[category] = round(negative_keywords[category] / count * 100, 1)
# 按出现次数排序
sorted_keywords = sorted(keyword_counts.items(), key=lambda x: x[1], reverse=True)
return {
"counts": dict(keyword_counts),
"sorted": sorted_keywords,
"negative_ratio": keyword_negative_ratio
}
def _analyze_variants(self, reviews: List[Dict]) -> Dict:
"""分析变体评论分布"""
variant_counts = defaultdict(int)
variant_ratings = defaultdict(list)
for review in reviews:
if review["variant_info"]:
variant = review["variant_info"]
variant_counts[variant] += 1
variant_ratings[variant].append(review["rating"])
# 计算每个变体的平均评分
variant_avg_rating = {}
for variant, ratings in variant_ratings.items():
variant_avg_rating[variant] = round(sum(ratings) / len(ratings), 1)
# 按评论数排序
sorted_variants = sorted(variant_counts.items(), key=lambda x: x[1], reverse=True)
return {
"counts": dict(variant_counts),
"avg_rating": variant_avg_rating,
"sorted": sorted_variants[:10] # 前10个热门变体
}
def _analyze_images(self, reviews: List[Dict]) -> Dict:
"""分析评论图片情况"""
has_image_count = 0
total_images = 0
image_reviews = []
for review in reviews:
if review["images"]:
has_image_count += 1
img_count = len(review["images"])
total_images += img_count
image_reviews.append({
"review_id": review["review_id"],
"image_count": img_count,
"rating": review["rating"],
"images": review["images"]
})
# 按图片数量排序
image_reviews_sorted = sorted(image_reviews, key=lambda x: x["image_count"], reverse=True)
return {
"has_image_ratio": round(has_image_count / len(reviews) * 100, 1) if reviews else 0,
"avg_images_per_review": round(total_images / len(reviews), 1) if reviews else 0,
"top_image_reviews": image_reviews_sorted[:5]
}
def _analyze_seller_replies(self, reviews: List[Dict]) -> Dict:
"""分析卖家回复情况"""
replied_count = 0
reply_times = []
for review in reviews:
if review["seller_reply"]:
replied_count += 1
# 计算回复时间（评论发布到卖家回复的时间差，单位：小时）
if review["created_time"] and review["seller_reply"]["created_time"]:
time_diff = (review["seller_reply"]["created_time"] - review["created_time"]) / 3600000
reply_times.append(time_diff)
avg_reply_time = round(sum(reply_times) / len(reply_times), 1) if reply_times else 0
return {
"reply_ratio": round(replied_count / len(reviews) * 100, 1) if reviews else 0,
"avg_reply_time_hours": avg_reply_time,
"replied_count": replied_count
}
def _get_typical_reviews(self, reviews: List[Dict]) -> Dict:
"""获取典型评论示例"""
# 按情感和评分分类的典型评论
typical = {
"positive_high_rating": self._get_top_reviews(reviews,
lambda x: x["sentiment"]["sentiment"] == "positive" and x["rating"] >= 4,
sort_key=lambda x: x["like_count"], reverse=True)[:3],
"negative_low_rating": self._get_top_reviews(reviews,
lambda x: x["sentiment"]["sentiment"] == "negative" and x["rating"] <= 2,
sort_key=lambda x: x["like_count"], reverse=True)[:3],
"most_helpful": self._get_top_reviews(reviews,
lambda x: True,
sort_key=lambda x: x["like_count"], reverse=True)[:3],
"with_images": self._get_top_reviews(reviews,
lambda x: len(x["images"]) > 0,
sort_key=lambda x: len(x["images"]), reverse=True)[:3]
}
return typical
def _get_top_reviews(self, reviews: List[Dict], filter_func, sort_key, reverse=True) -> List[Dict]:
"""筛选并排序评论"""
filtered = [r for r in reviews if filter_func(r)]
return sorted(filtered, key=sort_key, reverse=reverse)
def download_review_images(self, reviews: List[Dict], output_dir: str = "review_images") -> None:
"""下载评论图片"""
if not os.path.exists(output_dir):
os.makedirs(output_dir)
for i, review in enumerate(reviews):
if review["images"]:
for j, img_url in enumerate(review["images"]):
try:
response = self.session.get(img_url, timeout=10)
response.raise_for_status()
# 保存图片
img_path = f"{output_dir}/review_{review['review_id']}_img_{j}.jpg"
with open(img_path, 'wb') as f:
f.write(response.content)
logging.debug(f"已下载图片: {img_path}")
except Exception as e:
logging.error(f"下载图片失败 {img_url}: {str(e)}")
def visualize_analysis(self, analysis: Dict, output_dir: str = "review_analysis") -> None:
"""可视化分析结果"""
# 创建输出目录
if not os.path.exists(output_dir):
os.makedirs(output_dir)
# 设置中文显示
plt.rcParams["font.family"] = ["SimHei", "WenQuanYi Micro Hei", "Heiti TC"]
plt.rcParams["axes.unicode_minus"] = False
# 1. 评分分布条形图
plt.figure(figsize=(10, 6))
ratings = sorted(analysis["rating_analysis"]["counts"].items())
if ratings:
labels = [f"{k}星" for k, _ in ratings]
values = [v for _, v in ratings]
x = np.arange(len(labels))
plt.bar(x, values, color='skyblue')
plt.xticks(x, labels)
plt.ylabel('评论数量')
plt.title(f'商品评分分布 (平均评分: {analysis["rating_analysis"]["average"]})')
for i, v in enumerate(values):
plt.text(i, v + 1, str(v), ha='center')
plt.tight_layout()
plt.savefig(f"{output_dir}/rating_distribution.png")
plt.close()
# 2. 情感分布饼图
plt.figure(figsize=(8, 8))
sentiments = analysis["sentiment_analysis"]["distribution"]
if sentiments:
labels = ["正面", "负面", "中性"]
data = [
sentiments.get("positive", 0),
sentiments.get("negative", 0),
sentiments.get("neutral", 0)
]
colors = ['#4CAF50', '#F44336', '#FFC107']
plt.pie(data, labels=labels, colors=colors, autopct='%1.1f%%', startangle=90)
plt.title('评论情感分布')
plt.axis('equal')
plt.tight_layout()
plt.savefig(f"{output_dir}/sentiment_distribution.png")
plt.close()
# 3. 关键词分布条形图
plt.figure(figsize=(12, 6))
keywords = analysis["keyword_analysis"]["sorted"][:10]
if keywords:
labels = [k[0] for k in keywords]
values = [k[1] for k in keywords]
x = np.arange(len(labels))
plt.bar(x, values, color='#2196F3')
plt.xticks(x, labels)
plt.ylabel('出现次数')
plt.title('评论关键词分布')
plt.tight_layout()
plt.savefig(f"{output_dir}/keyword_distribution.png")
plt.close()
# 4. 变体评分对比图
plt.figure(figsize=(12, 6))
variants = analysis["variant_analysis"]["sorted"][:8]
if variants:
variant_names = [v[0][:15] + '...' if len(v[0]) > 15 else v[0] for v in variants]
variant_ratings = [analysis["variant_analysis"]["avg_rating"].get(v[0], 0) for v in variants]
x = np.arange(len(variant_names))
plt.bar(x, variant_ratings, color='#FF9800')
plt.xticks(x, variant_names, rotation=45)
plt.ylabel('平均评分')
plt.title('热门变体平均评分对比')
plt.ylim(0, 5)
for i, v in enumerate(variant_ratings):
plt.text(i, v + 0.1, str(v), ha='center')
plt.tight_layout()
plt.savefig(f"{output_dir}/variant_rating_comparison.png")
plt.close()
logging.info(f"分析图表已保存至 {output_dir} 目录")示例调用 if name == "main": # 替换为实际的参数（从Shopee开放平台获取） PARTNER_ID = 123456 # 合作伙伴ID PARTNER_KEY = "your_partner_key" # 合作伙伴密钥 SHOPID = 987654 # 店铺ID ACCESS_TOKEN = "your_access_token" # 访问令牌 ITEM_ID = 1234567890 # 商品ID REGION = "sg" # 地区代码
python
# 初始化API客户端
api = ShopeeItemReviewAPI(PARTNER_ID, PARTNER_KEY, SHOPID, region=REGION)
api.set_access_token(ACCESS_TOKEN)
# 获取商品评论
all_reviews = api.get_all_reviews(
item_id=ITEM_ID,
max_pages=5, # 最多获取5页
# rating_filter=5, # 只看5星评论
# with_images=True # 只看有图评论
)
if all_reviews:
# 分析评论数据
analysis = api.analyze_reviews(all_reviews)
print(f"=== Shopee商品评论分析报告 (商品ID: {ITEM_ID}) ===")
print(f"总评论数: {analysis['total_reviews']}条")
# 评分分析
print("\n评分分析:")
print(f" 平均评分: {analysis['rating_analysis']['average']}星")
print(" 评分分布:")
for rating, ratio in sorted(analysis['rating_analysis']['distribution'].items()):
print(f" {rating}星: {ratio}% ({analysis['rating_analysis']['counts'][rating]}条)")
# 情感分析
print("\n情感分析:")
print(f" 正面评论: {analysis['sentiment_analysis']['ratio'].get('positive', 0)}%")
print(f" 负面评论: {analysis['sentiment_analysis']['ratio'].get('negative', 0)}%")
print(f" 中性评论: {analysis['sentiment_analysis']['ratio'].get('neutral', 0)}%")
# 关键词分析
print("\n关键词分析:")
print(" 出现频率最高的5个关键词:")
for i, (keyword, count) in enumerate(analysis['keyword_analysis']['sorted'][:5], 1):
neg_ratio = analysis['keyword_analysis']['negative_ratio'].get(keyword, 0)
print(f" {i}. {keyword}: {count}次 (负面提及: {neg_ratio}%)")
# 变体分析
print("\n变体分析:")
if analysis['variant_analysis']['sorted']:
print(" 评论最多的3个变体:")
for i, (variant, count) in enumerate(analysis['variant_analysis']['sorted'][:3], 1):
rating = analysis['variant_analysis']['avg_rating'].get(variant, 0)
print(f" {i}. {variant}: {count}条评论, 平均评分: {rating}星")
# 图片分析
print("\n图片分析:")
print(f" 带图评论占比: {analysis['image_analysis']['has_image_ratio']}%")
print(f" 平均每条评论图片数: {analysis['image_analysis']['avg_images_per_review']}")
# 卖家回复分析
print("\n卖家回复分析:")
print(f" 回复率: {analysis['reply_analysis']['reply_ratio']}%")
print(f" 平均回复时间: {analysis['reply_analysis']['avg_reply_time_hours']}小时")
# 典型评论
print("\n典型正面评论:")
for i, review in enumerate(analysis['typical_reviews']['positive_high_rating'][:2], 1):
print(f" {i}. {review['user']['username']} ({review['rating_star']})")
print(f" 内容: {review['translated_content'][:100]}{'...' if len(review['translated_content'])>100 else ''}")
print("\n典型负面评论:")
for i, review in enumerate(analysis['typical_reviews']['negative_low_rating'][:2], 1):
print(f" {i}. {review['user']['username']} ({review['rating_star']})")
print(f" 内容: {review['translated_content'][:100]}{'...' if len(review['translated_content'])>100 else ''}")
# 生成可视化图表
api.visualize_analysis(analysis)
# 下载热门带图评论的图片
api.download_review_images(analysis['typical_reviews']['with_images'])三、接口调用注意事项
- 调用限制与规范
- QPS 限制：Shopee 开放平台对评论接口的 QPS 限制通常为 10 次 / 秒，不同地区可能有所差异
- 数据权限：只能获取自己店铺商品的评论数据，无法获取其他店铺的评论
- 多语言处理：Shopee 支持多地区多语言，需处理不同语言的评论内容
- 图片版权：评论图片属于用户所有，仅可用于店铺内部分析，不得用于商业宣传
- 合规使用：评论数据涉及用户隐私，需遵守各地区的数据保护法规
- 常见错误及解决方案
| 错误码 | 说明 | 解决方案 |
|---|---|---|
| 401 | 认证失败 | 检查 partner_id、partner_key 和 access_token 的有效性 |
| 403 | 权限不足 | 确认该店铺有访问评论数据的权限，检查 item_id 是否属于该店铺 |
| 404 | 商品不存在 | 确认 item_id 是否正确有效 |
| 429 | 调用频率超限 | 降低调用频率，实现请求限流 |
| 500 | 服务器错误 | 实现重试机制，最多 3 次，采用指数退避策略 |
| 10001 | 参数错误 | 检查请求参数格式，特别是 timestamp 和 sign 的生成是否正确 |
| 10003 | access_token 过期 | 重新获取 access_token |
- 数据解析要点
- 签名生成：Shopee 的签名机制较为复杂，需严格按照官方文档实现，注意时间戳的准确性
- 时间处理：评论时间为毫秒级时间戳，需转换为可读性强的日期格式
- 多语言处理：评论内容可能使用不同语言，需进行翻译后再分析
- 变体信息：变体属性可能以不同语言展示，需统一处理
- 图片 URL：评论图片 URL 可能有时效性，需及时下载或缓存
四、应用场景与扩展建议
典型应用场景
- 产品优化系统：基于评论反馈识别产品问题，指导产品改进
- 客户服务助手：自动识别负面评论并提醒客服及时回复
- 竞品分析工具：对比分析同类产品的用户评价和反馈
- 评论回复模板：基于常见问题生成回复模板，提高回复效率
- 产品描述优化：根据用户关注的关键词优化产品描述
扩展建议
- 实现实时监控：实时监测新评论，特别是负面评论，及时响应
- 开发自动分类系统：使用机器学习算法对评论进行自动分类和标签化
- 构建情感趋势分析：追踪情感变化趋势，评估产品改进效果
- 开发多维度对比功能：横向对比不同产品、不同时间段的评论特征
- 实现关键词预警：对出现频率激增的负面关键词进行预警
- 构建用户画像系统：基于评论内容分析用户特征和需求
通过合理使用 Shopee item_review 接口，卖家可以深入了解用户对产品的真实反馈，及时发现并解决问题，优化产品和服务，最终提升销售业绩和用户满意度。在实际应用中，需特别注意多语言处理和数据合规问题，确保符合各地区的法律法规要求。