企业如何建立豆包AI品牌提及监测体系

[问题库管理] → [自动化提问引擎] → [内容解析引擎] → [分析洞察平台] → [告警与报表]
      ↑                                                      ↓
      └──────────── [反馈优化闭环] ───────────────────────────┘

from playwright.sync_api import sync_playwright
import time
import json

class DoubaoMonitor:
    def __init__(self):
        self.playwright = sync_playwright().start()
        self.browser = self.playwright.chromium.launch(headless=False)
        self.context = self.browser.new_context(
            storage_state="doubao_auth.json"  # 登录态持久化
        )
        self.page = self.context.new_page()
    
    def login_doubao(self):
        """首次登录，保存登录态"""
        self.page.goto("https://www.doubao.com/chat/")
        input("请手动扫码登录后按回车继续...")
        self.context.storage_state(path="doubao_auth.json")
    
    def ask_question(self, question: str) -> dict:
        """向豆包提问并获取回答"""
        try:
            # 访问豆包对话页面
            self.page.goto("https://www.doubao.com/chat/")
            time.sleep(2)
            
            # 定位输入框并输入问题
            input_selector = "textarea[placeholder*='输入']"
            self.page.wait_for_selector(input_selector, timeout=10000)
            self.page.fill(input_selector, question)
            
            # 点击发送按钮
            send_selector = "button:has(svg)"  # 发送按钮通常有图标
            self.page.click(send_selector)
            
            # 等待AI回复生成完成
            time.sleep(10)  # 基础等待
            self.page.wait_for_selector(".answer-content", timeout=30000)
            
            # 提取回答内容
            answer_text = self.page.inner_text(".answer-content")
            
            # 提取引用来源（如有）
            sources = self.extract_sources()
            
            return {
                "question": question,
                "answer": answer_text,
                "sources": sources,
                "timestamp": time.time()
            }
        except Exception as e:
            return {
                "question": question,
                "answer": None,
                "error": str(e),
                "timestamp": time.time()
            }
    
    def extract_sources(self) -> list:
        """提取回答中引用的来源链接"""
        # 根据实际DOM结构调整选择器
        try:
            source_elements = self.page.query_selector_all(".source-link")
            return [el.inner_text() for el in source_elements]
        except:
            return []
    
    def batch_monitor(self, questions: list, interval: int = 30):
        """批量监测，注意频率控制"""
        results = []
        for i, q in enumerate(questions):
            print(f"正在提问 [{i+1}/{len(questions)}]: {q[:50]}...")
            result = self.ask_question(q)
            results.append(result)
            time.sleep(interval)  # 控制频率，避免触发反爬
        return results
    
    def close(self):
        self.browser.close()
        self.playwright.stop()

# 使用示例
if __name__ == "__main__":
    monitor = DoubaoMonitor()
    
    questions = [
        "XX品牌的面霜怎么样？",
        "XX品牌和YY品牌哪个好？",
        "XX品牌最近有负面新闻吗？"
    ]
    
    results = monitor.batch_monitor(questions, interval=30)
    
    # 保存结果
    with open("monitor_results.json", "w", encoding="utf-8") as f:
        json.dump(results, f, ensure_ascii=False, indent=2)
    
    monitor.close()

import openai  # 或其他LLM SDK

class AnswerAnalyzer:
    def __init__(self, api_key: str):
        self.client = openai.OpenAI(api_key=api_key)
    
    def analyze_sentiment(self, answer: str, brand_name: str) -> dict:
        """情感分析与关键信息提取"""
        prompt = f"""
        请分析以下AI助手的回答中，关于品牌"{brand_name}"的信息：
        
        回答内容：{answer}
        
        请按JSON格式返回：
        {{
            "sentiment": "positive/negative/neutral/mixed",
            "brand_mentioned": true/false,
            "competitors_mentioned": ["竞品1", "竞品2"],
            "key_points": ["要点1", "要点2"],
            "risk_signals": ["风险信号1"],
            "factual_errors": ["事实错误1"],
            "recommendation_tendency": "推荐/不推荐/中立/未提及",
            "summary": "一句话总结"
        }}
        """
        
        response = self.client.chat.completions.create(
            model="gpt-4o",
            messages=[{"role": "user", "content": prompt}],
            response_format={"type": "json_object"}
        )
        
        return json.loads(response.choices[0].message.content)
    
    def compare_answers(self, results: list) -> dict:
        """跨时间对比分析"""
        # 将多轮监测结果进行对比，发现变化趋势
        # 可以关注：情感变化、提及率变化、竞品对比变化
        pass

-- 核心监测数据表设计
CREATE TABLE brand_monitor_records (
    id BIGINT AUTO_INCREMENT PRIMARY KEY,
    brand_name VARCHAR(100) NOT NULL COMMENT '品牌名称',
    question_category VARCHAR(50) COMMENT '问题类别',
    question_text TEXT COMMENT '提问内容',
    answer_text MEDIUMTEXT COMMENT 'AI回答原文',
    sentiment VARCHAR(20) COMMENT '情感倾向',
    recommendation_tendency VARCHAR(20) COMMENT '推荐倾向',
    competitors_mentioned JSON COMMENT '提及的竞品',
    risk_signals JSON COMMENT '风险信号',
    sources JSON COMMENT '引用来源',
    is_error TINYINT DEFAULT 0 COMMENT '是否有事实错误',
    monitor_time DATETIME COMMENT '监测时间',
    created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
    INDEX idx_brand_time (brand_name, monitor_time),
    INDEX idx_sentiment (sentiment)
) COMMENT '品牌监测记录表';