419 lines
20 KiB
Python
419 lines
20 KiB
Python
|
"""
|
|||
|
|
配置优化助手模块
|
|||
|
|
分析品牌名和优势是否 GEO 友好，提供优化建议
|
|||
|
|
"""
|
|||
|
|
from typing import Dict, List, Optional
|
|||
|
|
from langchain_core.prompts import PromptTemplate
|
|||
|
|
from langchain_core.output_parsers import StrOutputParser
|
|||
|
|
import json
|
|||
|
|
import re
|
|||
|
|
|
|||
|
|
|
|||
|
|
class ConfigOptimizer:
|
|||
|
|
"""配置优化器"""
|
|||
|
|
|
|||
|
|
def __init__(self):
|
|||
|
|
self.optimization_prompt_template = """
|
|||
|
|
你是GEO（生成式引擎优化）专家，专注于帮助品牌在AI模型中被优先、可信地提及。
|
|||
|
|
|
|||
|
|
【当前配置】
|
|||
|
|
- 主品牌名称：{brand}
|
|||
|
|
- 核心优势/卖点：{advantages}
|
|||
|
|
- 竞品列表：{competitors}
|
|||
|
|
|
|||
|
|
【分析要求】
|
|||
|
|
请从以下维度全面评估当前配置，并给出优化建议：
|
|||
|
|
|
|||
|
|
1. **品牌名独特性分析**
|
|||
|
|
- 是否过于泛化（如"AI助手"、"智能系统"等通用词）？
|
|||
|
|
- 是否容易被混淆或误认为是其他品牌？
|
|||
|
|
- 是否具有搜索友好性（用户容易搜索到）？
|
|||
|
|
- 是否在AI回答中容易被识别和提及？
|
|||
|
|
|
|||
|
|
2. **优势描述分析**
|
|||
|
|
- 是否具体、可量化（避免"强大"、"优秀"等模糊词）？
|
|||
|
|
- 是否具有差异化（与竞品有明显区别）？
|
|||
|
|
- 是否包含E-E-A-T信号（专业性、经验性、权威性、可信度）？
|
|||
|
|
- 是否便于AI提取和引用？
|
|||
|
|
|
|||
|
|
3. **竞品对比分析**
|
|||
|
|
- 当前配置在竞品中是否具有明显优势？
|
|||
|
|
- 哪些方面容易被竞品超越？
|
|||
|
|
- 如何强化差异化定位？
|
|||
|
|
|
|||
|
|
4. **GEO友好度评估**
|
|||
|
|
- 品牌名是否容易被AI优先提及？
|
|||
|
|
- 优势描述是否符合GEO最佳实践？
|
|||
|
|
- 整体配置是否有助于提升提及率？
|
|||
|
|
|
|||
|
|
【输出格式】
|
|||
|
|
请严格按照以下格式输出，包含所有部分：
|
|||
|
|
|
|||
|
|
【评估总结】
|
|||
|
|
（200-300字，总结当前配置的优势和不足）
|
|||
|
|
|
|||
|
|
【优化建议】
|
|||
|
|
1. 品牌名优化建议：
|
|||
|
|
- 问题：[指出当前品牌名的问题]
|
|||
|
|
- 建议：[给出优化建议]
|
|||
|
|
|
|||
|
|
2. 优势描述优化建议：
|
|||
|
|
- 问题：[指出当前优势描述的问题]
|
|||
|
|
- 建议：[给出优化建议]
|
|||
|
|
|
|||
|
|
3. 差异化强化建议：
|
|||
|
|
- 竞品对比：[与竞品的对比分析]
|
|||
|
|
- 差异化策略：[如何强化差异化]
|
|||
|
|
|
|||
|
|
【推荐版本】
|
|||
|
|
请提供3个优化后的配置版本（从保守到激进），严格按照以下格式输出，每个字段单独一行：
|
|||
|
|
|
|||
|
|
版本1（保守优化）：
|
|||
|
|
品牌名：基于当前品牌名进行保守优化，保持核心品牌名不变
|
|||
|
|
优势描述：优化优势描述，使其更具体、可量化，用顿号分隔多个优势点
|
|||
|
|
理由：说明为什么这样优化，50-100字
|
|||
|
|
|
|||
|
|
版本2（平衡优化）：
|
|||
|
|
品牌名：在品牌名中加入行业关键词，提升搜索友好性
|
|||
|
|
优势描述：聚焦核心价值，突出差异化优势，用顿号分隔多个优势点
|
|||
|
|
理由：说明为什么这样优化，50-100字
|
|||
|
|
|
|||
|
|
版本3（激进优化）：
|
|||
|
|
品牌名：完全重构品牌定位，突出核心特性，最大化GEO效果
|
|||
|
|
优势描述：全面展示优势，包含多个维度，用顿号分隔多个优势点
|
|||
|
|
理由：说明为什么这样优化，50-100字
|
|||
|
|
|
|||
|
|
格式要求（非常重要，必须严格遵守）：
|
|||
|
|
1. 必须严格按照上述格式，每个字段单独一行，使用"品牌名："、"优势描述："、"理由："作为字段标识
|
|||
|
|
2. 品牌名必须提供实际内容，不能使用占位符，必须基于当前品牌名"{brand}"进行优化
|
|||
|
|
3. 优势描述必须提供实际内容，不能使用占位符，必须基于当前优势"{advantages}"进行优化
|
|||
|
|
4. 每个版本都必须完整，不能省略任何字段
|
|||
|
|
5. 不要使用方括号[]、不要使用占位符，必须提供针对当前配置的实际优化内容
|
|||
|
|
6. 品牌名和优势描述必须是具体的、可用的内容，不能是说明性文字
|
|||
|
|
|
|||
|
|
【预期效果】
|
|||
|
|
- 提及率提升预期：[预计提升幅度]
|
|||
|
|
- GEO友好度提升：[预计提升幅度]
|
|||
|
|
- 差异化优势：[预计强化效果]
|
|||
|
|
|
|||
|
|
【开始分析】
|
|||
|
|
"""
|
|||
|
|
|
|||
|
|
def optimize_config(self, brand: str, advantages: str, competitors: List[str], llm_chain) -> Dict:
|
|||
|
|
"""
|
|||
|
|
优化配置
|
|||
|
|
|
|||
|
|
Args:
|
|||
|
|
brand: 主品牌名称
|
|||
|
|
advantages: 核心优势/卖点
|
|||
|
|
competitors: 竞品列表
|
|||
|
|
llm_chain: LLM调用链
|
|||
|
|
|
|||
|
|
Returns:
|
|||
|
|
包含优化建议的字典
|
|||
|
|
"""
|
|||
|
|
competitors_str = "、".join(competitors) if competitors else "无"
|
|||
|
|
|
|||
|
|
prompt = PromptTemplate.from_template(self.optimization_prompt_template)
|
|||
|
|
chain = prompt | llm_chain | StrOutputParser()
|
|||
|
|
|
|||
|
|
try:
|
|||
|
|
result = chain.invoke({
|
|||
|
|
"brand": brand,
|
|||
|
|
"advantages": advantages,
|
|||
|
|
"competitors": competitors_str
|
|||
|
|
})
|
|||
|
|
|
|||
|
|
# 解析结果
|
|||
|
|
parsed_result = self._parse_optimization_result(result)
|
|||
|
|
parsed_result["raw_result"] = result
|
|||
|
|
parsed_result["success"] = True
|
|||
|
|
|
|||
|
|
# 如果推荐版本为空，尝试备用解析方法
|
|||
|
|
if not parsed_result.get("recommended_versions") or all(
|
|||
|
|
not v.get("brand") and not v.get("advantages")
|
|||
|
|
for v in parsed_result.get("recommended_versions", [])
|
|||
|
|
):
|
|||
|
|
# 尝试更宽松的解析
|
|||
|
|
parsed_result = self._parse_optimization_result_fallback(result, parsed_result)
|
|||
|
|
|
|||
|
|
return parsed_result
|
|||
|
|
except Exception as e:
|
|||
|
|
return {
|
|||
|
|
"success": False,
|
|||
|
|
"error": str(e),
|
|||
|
|
"raw_result": ""
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
def _parse_optimization_result(self, result: str) -> Dict:
|
|||
|
|
"""解析优化结果"""
|
|||
|
|
parsed = {
|
|||
|
|
"summary": "",
|
|||
|
|
"suggestions": {
|
|||
|
|
"brand": {"problem": "", "suggestion": ""},
|
|||
|
|
"advantages": {"problem": "", "suggestion": ""},
|
|||
|
|
"differentiation": {"comparison": "", "strategy": ""}
|
|||
|
|
},
|
|||
|
|
"recommended_versions": [],
|
|||
|
|
"expected_effects": {},
|
|||
|
|
"parse_errors": [] # 记录解析过程中的错误
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
# 提取评估总结
|
|||
|
|
if "【评估总结】" in result:
|
|||
|
|
summary_section = result.split("【评估总结】")[1].split("【")[0].strip()
|
|||
|
|
parsed["summary"] = summary_section
|
|||
|
|
|
|||
|
|
# 提取优化建议
|
|||
|
|
if "【优化建议】" in result:
|
|||
|
|
suggestions_section = result.split("【优化建议】")[1].split("【")[0]
|
|||
|
|
|
|||
|
|
# 品牌名优化建议 - 更健壮的解析
|
|||
|
|
brand_patterns = ["品牌名优化建议", "1. 品牌名优化建议", "品牌名"]
|
|||
|
|
for pattern in brand_patterns:
|
|||
|
|
if pattern in suggestions_section:
|
|||
|
|
brand_section = suggestions_section.split(pattern, 1)[1]
|
|||
|
|
# 找到下一个建议的开始位置
|
|||
|
|
next_patterns = ["2. 优势描述优化建议", "优势描述优化建议", "3. 差异化强化建议"]
|
|||
|
|
for next_pattern in next_patterns:
|
|||
|
|
if next_pattern in brand_section:
|
|||
|
|
brand_section = brand_section.split(next_pattern)[0]
|
|||
|
|
break
|
|||
|
|
|
|||
|
|
if "问题：" in brand_section or "问题" in brand_section:
|
|||
|
|
problem_text = brand_section.split("问题：")[1] if "问题：" in brand_section else brand_section.split("问题")[1]
|
|||
|
|
problem_text = problem_text.split("建议：")[0].split("建议")[0].strip()
|
|||
|
|
if problem_text:
|
|||
|
|
parsed["suggestions"]["brand"]["problem"] = problem_text
|
|||
|
|
|
|||
|
|
if "建议：" in brand_section or "建议" in brand_section:
|
|||
|
|
suggestion_text = brand_section.split("建议：")[1] if "建议：" in brand_section else brand_section.split("建议", 1)[1]
|
|||
|
|
suggestion_text = suggestion_text.split("2.")[0].split("3.")[0].strip()
|
|||
|
|
if suggestion_text:
|
|||
|
|
parsed["suggestions"]["brand"]["suggestion"] = suggestion_text
|
|||
|
|
break
|
|||
|
|
|
|||
|
|
# 优势描述优化建议 - 更健壮的解析
|
|||
|
|
adv_patterns = ["优势描述优化建议", "2. 优势描述优化建议", "优势描述"]
|
|||
|
|
for pattern in adv_patterns:
|
|||
|
|
if pattern in suggestions_section:
|
|||
|
|
adv_section = suggestions_section.split(pattern, 1)[1]
|
|||
|
|
# 找到下一个建议的开始位置
|
|||
|
|
next_patterns = ["3. 差异化强化建议", "差异化强化建议"]
|
|||
|
|
for next_pattern in next_patterns:
|
|||
|
|
if next_pattern in adv_section:
|
|||
|
|
adv_section = adv_section.split(next_pattern)[0]
|
|||
|
|
break
|
|||
|
|
|
|||
|
|
if "问题：" in adv_section or "问题" in adv_section:
|
|||
|
|
problem_text = adv_section.split("问题：")[1] if "问题：" in adv_section else adv_section.split("问题")[1]
|
|||
|
|
problem_text = problem_text.split("建议：")[0].split("建议")[0].strip()
|
|||
|
|
if problem_text:
|
|||
|
|
parsed["suggestions"]["advantages"]["problem"] = problem_text
|
|||
|
|
|
|||
|
|
if "建议：" in adv_section or "建议" in adv_section:
|
|||
|
|
suggestion_text = adv_section.split("建议：")[1] if "建议：" in adv_section else adv_section.split("建议", 1)[1]
|
|||
|
|
suggestion_text = suggestion_text.split("3.")[0].strip()
|
|||
|
|
if suggestion_text:
|
|||
|
|
parsed["suggestions"]["advantages"]["suggestion"] = suggestion_text
|
|||
|
|
break
|
|||
|
|
|
|||
|
|
# 提取推荐版本
|
|||
|
|
if "【推荐版本】" in result:
|
|||
|
|
versions_section = result.split("【推荐版本】")[1].split("【")[0]
|
|||
|
|
|
|||
|
|
# 提取3个版本 - 使用更健壮的解析方式
|
|||
|
|
for i in range(1, 4):
|
|||
|
|
# 尝试多种匹配模式
|
|||
|
|
version_patterns = [
|
|||
|
|
f"版本{i}（",
|
|||
|
|
f"版本{i}：",
|
|||
|
|
f"版本{i}",
|
|||
|
|
f"版本 {i}",
|
|||
|
|
f"Version {i}",
|
|||
|
|
]
|
|||
|
|
|
|||
|
|
version_text = None
|
|||
|
|
for pattern in version_patterns:
|
|||
|
|
if pattern in versions_section:
|
|||
|
|
# 找到版本开始位置
|
|||
|
|
start_idx = versions_section.find(pattern)
|
|||
|
|
version_text = versions_section[start_idx + len(pattern):]
|
|||
|
|
|
|||
|
|
# 找到下一个版本或结束位置
|
|||
|
|
next_patterns = [
|
|||
|
|
f"版本{i+1}（" if i < 3 else None,
|
|||
|
|
f"版本{i+1}：" if i < 3 else None,
|
|||
|
|
f"版本{i+1}" if i < 3 else None,
|
|||
|
|
"【预期效果】",
|
|||
|
|
"【",
|
|||
|
|
]
|
|||
|
|
|
|||
|
|
end_idx = len(version_text)
|
|||
|
|
for next_pattern in next_patterns:
|
|||
|
|
if next_pattern and next_pattern in version_text:
|
|||
|
|
end_idx = min(end_idx, version_text.find(next_pattern))
|
|||
|
|
|
|||
|
|
version_text = version_text[:end_idx].strip()
|
|||
|
|
break
|
|||
|
|
|
|||
|
|
if not version_text:
|
|||
|
|
continue
|
|||
|
|
|
|||
|
|
version_data = {
|
|||
|
|
"version_name": f"版本{i}",
|
|||
|
|
"brand": "",
|
|||
|
|
"advantages": "",
|
|||
|
|
"reason": ""
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
# 提取品牌名 - 支持多种格式，更健壮的解析
|
|||
|
|
brand_patterns = ["品牌名：", "品牌名", "品牌：", "- 品牌名：", "品牌名"]
|
|||
|
|
for pattern in brand_patterns:
|
|||
|
|
if pattern in version_text:
|
|||
|
|
brand_part = version_text.split(pattern, 1)[1]
|
|||
|
|
# 提取到换行、下一个字段或下一个版本
|
|||
|
|
brand = brand_part.split("\n")[0]
|
|||
|
|
# 移除可能的下一个字段标识
|
|||
|
|
for next_field in ["优势描述", "理由", "版本", "- 优势描述", "- 理由"]:
|
|||
|
|
if next_field in brand:
|
|||
|
|
brand = brand.split(next_field)[0]
|
|||
|
|
brand = brand.strip()
|
|||
|
|
# 移除可能的冒号、破折号等
|
|||
|
|
brand = brand.lstrip("：").lstrip(":").lstrip("-").lstrip("—").strip()
|
|||
|
|
if brand and len(brand) > 0:
|
|||
|
|
version_data["brand"] = brand
|
|||
|
|
break
|
|||
|
|
|
|||
|
|
# 提取优势描述 - 支持多种格式，更健壮的解析
|
|||
|
|
adv_patterns = ["优势描述：", "优势描述", "优势：", "- 优势描述："]
|
|||
|
|
for pattern in adv_patterns:
|
|||
|
|
if pattern in version_text:
|
|||
|
|
adv_part = version_text.split(pattern, 1)[1]
|
|||
|
|
# 提取到理由或下一个字段
|
|||
|
|
advantages = adv_part.split("理由")[0].split("- 理由")[0].split("版本")[0]
|
|||
|
|
# 处理多行内容
|
|||
|
|
advantages_lines = []
|
|||
|
|
for line in advantages.split("\n"):
|
|||
|
|
line = line.strip()
|
|||
|
|
# 如果遇到下一个字段标识，停止
|
|||
|
|
if any(marker in line for marker in ["理由", "版本", "品牌名"]):
|
|||
|
|
break
|
|||
|
|
if line and not line.startswith("-") and not line.startswith("—"):
|
|||
|
|
advantages_lines.append(line)
|
|||
|
|
|
|||
|
|
advantages = " ".join(advantages_lines).strip()
|
|||
|
|
# 移除可能的冒号、破折号等
|
|||
|
|
advantages = advantages.lstrip("：").lstrip(":").lstrip("-").lstrip("—").strip()
|
|||
|
|
if advantages and len(advantages) > 0:
|
|||
|
|
version_data["advantages"] = advantages
|
|||
|
|
break
|
|||
|
|
|
|||
|
|
# 提取理由 - 支持多种格式，更健壮的解析
|
|||
|
|
reason_patterns = ["理由：", "理由", "- 理由："]
|
|||
|
|
for pattern in reason_patterns:
|
|||
|
|
if pattern in version_text:
|
|||
|
|
reason_part = version_text.split(pattern, 1)[1]
|
|||
|
|
# 提取到下一个版本或结束
|
|||
|
|
reason = reason_part.split("版本")[0]
|
|||
|
|
# 处理多行内容
|
|||
|
|
reason_lines = []
|
|||
|
|
for line in reason.split("\n"):
|
|||
|
|
line = line.strip()
|
|||
|
|
# 如果遇到下一个版本标识，停止
|
|||
|
|
if "版本" in line and any(str(i+1) in line for i in range(1, 4)):
|
|||
|
|
break
|
|||
|
|
if line and not line.startswith("-") and not line.startswith("—"):
|
|||
|
|
reason_lines.append(line)
|
|||
|
|
|
|||
|
|
reason = " ".join(reason_lines).strip()
|
|||
|
|
# 移除可能的冒号、破折号等
|
|||
|
|
reason = reason.lstrip("：").lstrip(":").lstrip("-").lstrip("—").strip()
|
|||
|
|
if reason and len(reason) > 0:
|
|||
|
|
version_data["reason"] = reason
|
|||
|
|
break
|
|||
|
|
|
|||
|
|
# 只有当至少有一个字段有内容时才添加
|
|||
|
|
if version_data["brand"] or version_data["advantages"]:
|
|||
|
|
parsed["recommended_versions"].append(version_data)
|
|||
|
|
else:
|
|||
|
|
# 记录解析失败的版本
|
|||
|
|
parsed["parse_errors"].append(f"版本{i}解析失败：未找到品牌名或优势描述")
|
|||
|
|
|
|||
|
|
# 提取预期效果
|
|||
|
|
if "【预期效果】" in result:
|
|||
|
|
effects_section = result.split("【预期效果】")[1].strip()
|
|||
|
|
if "提及率提升预期：" in effects_section:
|
|||
|
|
parsed["expected_effects"]["mention_rate"] = effects_section.split("提及率提升预期：")[1].split("\n")[0].strip()
|
|||
|
|
if "GEO友好度提升：" in effects_section:
|
|||
|
|
parsed["expected_effects"]["geo_friendliness"] = effects_section.split("GEO友好度提升：")[1].split("\n")[0].strip()
|
|||
|
|
|
|||
|
|
return parsed
|
|||
|
|
|
|||
|
|
def _parse_optimization_result_fallback(self, result: str, parsed: Dict) -> Dict:
|
|||
|
|
"""备用解析方法，使用更宽松的规则和正则表达式"""
|
|||
|
|
# 如果推荐版本部分存在但解析失败，尝试更宽松的解析
|
|||
|
|
if "【推荐版本】" in result:
|
|||
|
|
versions_section = result.split("【推荐版本】")[1].split("【")[0]
|
|||
|
|
|
|||
|
|
# 使用正则表达式提取
|
|||
|
|
import re
|
|||
|
|
|
|||
|
|
# 尝试提取版本1
|
|||
|
|
version1_match = re.search(r'版本1[^版本]*?品牌名[：:]\s*([^\n]+)', versions_section, re.DOTALL)
|
|||
|
|
version1_adv_match = re.search(r'版本1[^版本]*?优势描述[：:]\s*([^\n]+)', versions_section, re.DOTALL)
|
|||
|
|
|
|||
|
|
if version1_match or version1_adv_match:
|
|||
|
|
v1 = {
|
|||
|
|
"version_name": "版本1",
|
|||
|
|
"brand": version1_match.group(1).strip() if version1_match else "",
|
|||
|
|
"advantages": version1_adv_match.group(1).strip() if version1_adv_match else "",
|
|||
|
|
"reason": ""
|
|||
|
|
}
|
|||
|
|
# 清理品牌名和优势描述
|
|||
|
|
v1["brand"] = v1["brand"].split("优势描述")[0].split("理由")[0].strip()
|
|||
|
|
v1["advantages"] = v1["advantages"].split("理由")[0].split("版本")[0].strip()
|
|||
|
|
if v1["brand"] or v1["advantages"]:
|
|||
|
|
if not parsed["recommended_versions"]:
|
|||
|
|
parsed["recommended_versions"] = []
|
|||
|
|
if len(parsed["recommended_versions"]) < 1:
|
|||
|
|
parsed["recommended_versions"].append(v1)
|
|||
|
|
|
|||
|
|
# 尝试提取版本2
|
|||
|
|
version2_match = re.search(r'版本2[^版本]*?品牌名[：:]\s*([^\n]+)', versions_section, re.DOTALL)
|
|||
|
|
version2_adv_match = re.search(r'版本2[^版本]*?优势描述[：:]\s*([^\n]+)', versions_section, re.DOTALL)
|
|||
|
|
|
|||
|
|
if version2_match or version2_adv_match:
|
|||
|
|
v2 = {
|
|||
|
|
"version_name": "版本2",
|
|||
|
|
"brand": version2_match.group(1).strip() if version2_match else "",
|
|||
|
|
"advantages": version2_adv_match.group(1).strip() if version2_adv_match else "",
|
|||
|
|
"reason": ""
|
|||
|
|
}
|
|||
|
|
# 清理品牌名和优势描述
|
|||
|
|
v2["brand"] = v2["brand"].split("优势描述")[0].split("理由")[0].strip()
|
|||
|
|
v2["advantages"] = v2["advantages"].split("理由")[0].split("版本")[0].strip()
|
|||
|
|
if v2["brand"] or v2["advantages"]:
|
|||
|
|
if len(parsed["recommended_versions"]) < 2:
|
|||
|
|
parsed["recommended_versions"].append(v2)
|
|||
|
|
|
|||
|
|
# 尝试提取版本3
|
|||
|
|
version3_match = re.search(r'版本3[^版本]*?品牌名[：:]\s*([^\n]+)', versions_section, re.DOTALL)
|
|||
|
|
version3_adv_match = re.search(r'版本3[^版本]*?优势描述[：:]\s*([^\n]+)', versions_section, re.DOTALL)
|
|||
|
|
|
|||
|
|
if version3_match or version3_adv_match:
|
|||
|
|
v3 = {
|
|||
|
|
"version_name": "版本3",
|
|||
|
|
"brand": version3_match.group(1).strip() if version3_match else "",
|
|||
|
|
"advantages": version3_adv_match.group(1).strip() if version3_adv_match else "",
|
|||
|
|
"reason": ""
|
|||
|
|
}
|
|||
|
|
# 清理品牌名和优势描述
|
|||
|
|
v3["brand"] = v3["brand"].split("优势描述")[0].split("理由")[0].strip()
|
|||
|
|
v3["advantages"] = v3["advantages"].split("理由")[0].split("版本")[0].strip()
|
|||
|
|
if v3["brand"] or v3["advantages"]:
|
|||
|
|
if len(parsed["recommended_versions"]) < 3:
|
|||
|
|
parsed["recommended_versions"].append(v3)
|
|||
|
|
|
|||
|
|
return parsed
|