GEO自动化工具链
从监测到优化的全链路自动化

# AI搜索监测自建脚本 (Playwright + Python)

from playwright.sync_api import sync_playwright
import json, time
from datetime import datetime

PLATFORMS = {
    'chatgpt': {'url': 'https://chatgpt.com', 'input_selector': 'textarea'},
    'perplexity': {'url': 'https://perplexity.ai', 'input_selector': 'textarea'},
    'kimi': {'url': 'https://kimi.moonshot.cn', 'input_selector': 'textarea'}
}

KEYWORDS = [
    '企业级项目管理软件推荐',
    '最佳SaaS项目管理工具',
    '项目管理软件对比2026'
]

BRAND_NAME = 'XX科技'

def monitor_ai_search(platform, keyword):
    """在指定AI平台搜索关键词并提取品牌引用信息"""
    with sync_playwright() as p:
        browser = p.chromium.launch(headless=True)
        page = browser.new_page()
        
        config = PLATFORMS[platform]
        page.goto(config['url'])
        time.sleep(3)
        
        page.fill(config['input_selector'], keyword)
        page.press(config['input_selector'], 'Enter')
        time.sleep(15)
        
        response_text = page.inner_text('.response-container')
        
        result = {
            'platform': platform,
            'keyword': keyword,
            'timestamp': datetime.now().isoformat(),
            'brand_mentioned': BRAND_NAME in response_text,
            'response_length': len(response_text),
            'response_excerpt': response_text[:500]
        }
        
        browser.close()
        return result

results = []
for platform in PLATFORMS:
    for keyword in KEYWORDS:
        result = monitor_ai_search(platform, keyword)
        results.append(result)
        time.sleep(5)

with open(f'geo_monitor_{datetime.now().strftime("%Y%m%d")}.json', 'w', encoding='utf-8') as f:
    json.dump(results, f, ensure_ascii=False, indent=2)

# GEO效果分析引擎

import statistics
from collections import defaultdict

class GEOAnalyticsEngine:
    """GEO效果分析引擎"""
    
    def __init__(self, monitoring_data):
        self.data = monitoring_data
    
    def calculate_citation_rate(self, period='7d'):
        """计算AI引用率"""
        total_queries = len(self.data)
        cited_queries = sum(1 for d in self.data if d['brand_mentioned'])
        return cited_queries / total_queries * 100 if total_queries > 0 else 0
    
    def calculate_brand_accuracy(self):
        """计算品牌描述准确度"""
        mentioned = [d for d in self.data if d['brand_mentioned']]
        if not mentioned:
            return 0
        correct = sum(1 for d in mentioned if d.get('description_accurate', False))
        return correct / len(mentioned) * 100
    
    def trend_analysis(self, metric='citation_rate', window=7):
        """趋势分析 - 计算移动平均"""
        daily_rates = defaultdict(list)
        for d in self.data:
            date = d['timestamp'][:10]
            daily_rates[date].append(d['brand_mentioned'])
        
        daily_avg = {}
        for date, mentions in sorted(daily_rates.items()):
            daily_avg[date] = sum(mentions) / len(mentions) * 100
        
        dates = list(daily_avg.keys())
        ma_values = []
        for i in range(window - 1, len(dates)):
            window_vals = [daily_avg[dates[j]] for j in range(i - window + 1, i + 1)]
            ma_values.append({
                'date': dates[i],
                'value': statistics.mean(window_vals),
                'trend': 'up' if window_vals[-1] > window_vals[0] else 'down'
            })
        return ma_values
    
    def competitor_analysis(self, competitors):
        """竞品分析"""
        results = {}
        for comp in competitors:
            comp_mentions = sum(1 for d in self.data if comp in d.get('response_excerpt', ''))
            results[comp] = {
                'mentions': comp_mentions,
                'mention_rate': comp_mentions / len(self.data) * 100,
                'co_occurrence': sum(1 for d in self.data 
                    if comp in d.get('response_excerpt', '') and d['brand_mentioned'])
            }
        return results

# Schema标记自动生成工具

import json

def generate_article_schema(title, author, date_published, date_modified, 
                           description, keywords, url, image_url=None):
    """生成Article Schema JSON-LD"""
    schema = {
        '@context': 'https://schema.org',
        '@type': 'Article',
        'headline': title,
        'description': description,
        'keywords': keywords,
        'author': {'@type': 'Person', 'name': author},
        'datePublished': date_published,
        'dateModified': date_modified,
        'url': url,
        'publisher': {
            '@type': 'Organization',
            'name': 'XX科技',
            'logo': {'@type': 'ImageObject', 'url': 'https://example.com/logo.png'}
        }
    }
    if image_url:
        schema['image'] = image_url
    return json.dumps(schema, ensure_ascii=False, indent=2)

def generate_faq_schema(faq_list):
    """生成FAQPage Schema"""
    entities = []
    for item in faq_list:
        entities.append({
            '@type': 'Question',
            'name': item['question'],
            'acceptedAnswer': {'@type': 'Answer', 'text': item['answer']}
        })
    schema = {
        '@context': 'https://schema.org',
        '@type': 'FAQPage',
        'mainEntity': entities
    }
    return json.dumps(schema, ensure_ascii=False, indent=2)

def validate_content_quality(content, min_words=1800):
    """内容质量自动检查"""
    import re
    issues = []
    word_count = len(content)
    if word_count < min_words:
        issues.append(f'字数不足: {word_count} < {min_words}')
    
    numbers = len(re.findall(r'\d+', content))
    expected_numbers = word_count / 300
    if numbers < expected_numbers:
        issues.append(f'数据密度不足: {numbers}个数字, 期望{expected_numbers:.0f}个')
    
    if '## ' not in content:
        issues.append('缺少H2标题层级')
    
    return {'passed': len(issues) == 0, 'issues': issues}

# GEO工具链集成架构

toolchain_config = {
    'data_collection': {
        'tools': ['Profound API', 'Goodie API', 'Playwright脚本', '服务器日志分析'],
        'schedule': '每日2次品牌词 + 每日1次行业词',
        'storage': 'PostgreSQL',
        'retention': '12个月'
    },
    'analysis': {
        'tools': ['Python + Pandas', 'GEOAnalyticsEngine'],
        'schedule': '每日计算 + 每周深度分析',
        'outputs': ['指标仪表板数据', '趋势告警', '竞品分析报告']
    },
    'content_management': {
        'tools': ['GPT-4 API', 'Schema生成器', '内容质量检查器'],
        'workflow': 'AI生成大纲 -> 人工审核 -> AI初稿 -> 人工修改 -> 质量检查 -> 发布',
        'automation': '70%自动化, 30%人工'
    },
    'technical_monitoring': {
        'tools': ['Schema验证脚本', 'PageSpeed API', 'robots.txt监控', '爬虫日志分析'],
        'schedule': '每日检查 + 每周深度扫描',
        'auto_fix': '60%问题自动修复'
    },
    'reporting': {
        'tools': ['Python + Jinja2', 'Grafana看板', '飞书/钉钉推送'],
        'schedule': '周报(每周一) + 月报(每月第一周)',
        'audience': ['GEO团队', '市场总监', 'CTO']
    }
}