528 lines
19 KiB
Python
528 lines
19 KiB
Python
|
"""
|
|||
|
|
智能关键词挖掘与趋势分析模块
|
|||
|
|
支持行业热点挖掘、竞争度分析、趋势预测、价值矩阵等功能
|
|||
|
|
"""
|
|||
|
|
import pandas as pd
|
|||
|
|
from typing import List, Dict, Optional, Any, Tuple
|
|||
|
|
from datetime import datetime, timedelta
|
|||
|
|
import json
|
|||
|
|
from collections import defaultdict
|
|||
|
|
|
|||
|
|
|
|||
|
|
class KeywordMining:
|
|||
|
|
"""关键词挖掘与趋势分析引擎"""
|
|||
|
|
|
|||
|
|
def __init__(self, storage):
|
|||
|
|
"""
|
|||
|
|
Args:
|
|||
|
|
storage: DataStorage 实例
|
|||
|
|
"""
|
|||
|
|
self.storage = storage
|
|||
|
|
|
|||
|
|
def mine_industry_keywords(
|
|||
|
|
self,
|
|||
|
|
brand: str,
|
|||
|
|
industry: str,
|
|||
|
|
advantages: str,
|
|||
|
|
num_keywords: int = 20,
|
|||
|
|
llm_chain=None
|
|||
|
|
) -> List[Dict[str, Any]]:
|
|||
|
|
"""
|
|||
|
|
挖掘行业热点关键词
|
|||
|
|
|
|||
|
|
Args:
|
|||
|
|
brand: 品牌名称
|
|||
|
|
industry: 行业领域
|
|||
|
|
advantages: 品牌优势
|
|||
|
|
num_keywords: 需要挖掘的关键词数量
|
|||
|
|
llm_chain: LLM 调用链（可选）
|
|||
|
|
|
|||
|
|
Returns:
|
|||
|
|
关键词列表，每个关键词包含：
|
|||
|
|
- keyword: 关键词文本
|
|||
|
|
- category: 关键词类别
|
|||
|
|
- intent: 用户搜索意图
|
|||
|
|
- estimated_value: 预估价值
|
|||
|
|
"""
|
|||
|
|
if not llm_chain:
|
|||
|
|
# 如果没有 LLM，返回空列表
|
|||
|
|
return []
|
|||
|
|
|
|||
|
|
prompt = f"""你是关键词挖掘专家，专注于发现高价值的行业关键词。
|
|||
|
|
|
|||
|
|
【品牌信息】
|
|||
|
|
- 品牌：{brand}
|
|||
|
|
- 行业：{industry}
|
|||
|
|
- 核心优势：{advantages}
|
|||
|
|
|
|||
|
|
【任务】
|
|||
|
|
挖掘 {num_keywords} 个高价值关键词，这些关键词应该：
|
|||
|
|
1. 符合用户真实搜索意图
|
|||
|
|
2. 与品牌和行业高度相关
|
|||
|
|
3. 具有商业价值（用户有购买/使用意向）
|
|||
|
|
4. 覆盖不同搜索意图（对比、评测、使用、购买等）
|
|||
|
|
|
|||
|
|
【输出格式】
|
|||
|
|
请以 JSON 数组格式输出，每个关键词包含：
|
|||
|
|
- keyword: 关键词文本（12-28字）
|
|||
|
|
- category: 关键词类别（如：对比、评测、使用、购买、问题等）
|
|||
|
|
- intent: 用户搜索意图（如：了解产品、对比选择、使用教程等）
|
|||
|
|
- estimated_value: 预估价值（1-10分，10分最高）
|
|||
|
|
|
|||
|
|
【示例】
|
|||
|
|
[
|
|||
|
|
{{
|
|||
|
|
"keyword": "最好的{industry}软件有哪些",
|
|||
|
|
"category": "对比",
|
|||
|
|
"intent": "对比选择",
|
|||
|
|
"estimated_value": 9
|
|||
|
|
}},
|
|||
|
|
{{
|
|||
|
|
"keyword": "{brand}使用教程",
|
|||
|
|
"category": "使用",
|
|||
|
|
"intent": "使用教程",
|
|||
|
|
"estimated_value": 8
|
|||
|
|
}}
|
|||
|
|
]
|
|||
|
|
|
|||
|
|
【开始输出 JSON 数组】
|
|||
|
|
"""
|
|||
|
|
|
|||
|
|
try:
|
|||
|
|
result = llm_chain.invoke({"input": prompt})
|
|||
|
|
|
|||
|
|
# 尝试解析 JSON
|
|||
|
|
if isinstance(result, str):
|
|||
|
|
# 尝试提取 JSON 数组
|
|||
|
|
import re
|
|||
|
|
json_match = re.search(r'\[[\s\S]*\]', result)
|
|||
|
|
if json_match:
|
|||
|
|
keywords = json.loads(json_match.group(0))
|
|||
|
|
else:
|
|||
|
|
keywords = []
|
|||
|
|
else:
|
|||
|
|
keywords = result if isinstance(result, list) else []
|
|||
|
|
|
|||
|
|
# 验证和清理数据
|
|||
|
|
cleaned_keywords = []
|
|||
|
|
for kw in keywords:
|
|||
|
|
if isinstance(kw, dict) and "keyword" in kw:
|
|||
|
|
cleaned_keywords.append({
|
|||
|
|
"keyword": kw.get("keyword", ""),
|
|||
|
|
"category": kw.get("category", "其他"),
|
|||
|
|
"intent": kw.get("intent", "未知"),
|
|||
|
|
"estimated_value": kw.get("estimated_value", 5)
|
|||
|
|
})
|
|||
|
|
|
|||
|
|
return cleaned_keywords[:num_keywords]
|
|||
|
|
|
|||
|
|
except Exception as e:
|
|||
|
|
print(f"关键词挖掘失败: {e}")
|
|||
|
|
return []
|
|||
|
|
|
|||
|
|
def analyze_competition(
|
|||
|
|
self,
|
|||
|
|
keywords: List[str],
|
|||
|
|
brand: str,
|
|||
|
|
verify_llms: Dict = None
|
|||
|
|
) -> Dict[str, Dict[str, Any]]:
|
|||
|
|
"""
|
|||
|
|
分析关键词竞争度（在 AI 中的提及频率）
|
|||
|
|
|
|||
|
|
Args:
|
|||
|
|
keywords: 关键词列表
|
|||
|
|
brand: 品牌名称
|
|||
|
|
verify_llms: 验证用的 LLM 字典（可选，用于实时验证）
|
|||
|
|
|
|||
|
|
Returns:
|
|||
|
|
每个关键词的竞争度分析结果：
|
|||
|
|
- mention_rate: 提及率（0-1）
|
|||
|
|
- competition_level: 竞争级别（低/中/高）
|
|||
|
|
- competitor_mentions: 竞品提及次数
|
|||
|
|
- total_mentions: 总提及次数
|
|||
|
|
"""
|
|||
|
|
# 获取历史验证数据
|
|||
|
|
verify_df = self.storage.get_verify_results(brand=brand, include_timestamp=True)
|
|||
|
|
|
|||
|
|
competition_analysis = {}
|
|||
|
|
|
|||
|
|
for keyword in keywords:
|
|||
|
|
# 从历史数据中查找该关键词的验证结果
|
|||
|
|
keyword_verifications = verify_df[verify_df["问题"] == keyword] if not verify_df.empty else pd.DataFrame()
|
|||
|
|
|
|||
|
|
if not keyword_verifications.empty:
|
|||
|
|
# 计算提及率
|
|||
|
|
total_checks = len(keyword_verifications)
|
|||
|
|
brand_mentions = len(keyword_verifications[keyword_verifications["品牌"] == brand])
|
|||
|
|
mention_rate = brand_mentions / total_checks if total_checks > 0 else 0
|
|||
|
|
|
|||
|
|
# 计算竞品提及次数
|
|||
|
|
competitor_mentions = len(keyword_verifications[keyword_verifications["品牌"] != brand])
|
|||
|
|
|
|||
|
|
# 计算总提及次数（所有品牌）
|
|||
|
|
total_mentions = keyword_verifications["提及次数"].sum()
|
|||
|
|
|
|||
|
|
# 判断竞争级别
|
|||
|
|
if mention_rate < 0.3:
|
|||
|
|
competition_level = "高"
|
|||
|
|
elif mention_rate < 0.6:
|
|||
|
|
competition_level = "中"
|
|||
|
|
else:
|
|||
|
|
competition_level = "低"
|
|||
|
|
|
|||
|
|
competition_analysis[keyword] = {
|
|||
|
|
"mention_rate": mention_rate,
|
|||
|
|
"competition_level": competition_level,
|
|||
|
|
"competitor_mentions": int(competitor_mentions),
|
|||
|
|
"total_mentions": int(total_mentions),
|
|||
|
|
"data_points": total_checks
|
|||
|
|
}
|
|||
|
|
else:
|
|||
|
|
# 如果没有历史数据，返回默认值
|
|||
|
|
competition_analysis[keyword] = {
|
|||
|
|
"mention_rate": 0.0,
|
|||
|
|
"competition_level": "未知",
|
|||
|
|
"competitor_mentions": 0,
|
|||
|
|
"total_mentions": 0,
|
|||
|
|
"data_points": 0
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return competition_analysis
|
|||
|
|
|
|||
|
|
def predict_trend(
|
|||
|
|
self,
|
|||
|
|
keywords: List[str],
|
|||
|
|
brand: str,
|
|||
|
|
days: int = 30
|
|||
|
|
) -> Dict[str, Dict[str, Any]]:
|
|||
|
|
"""
|
|||
|
|
预测关键词趋势
|
|||
|
|
|
|||
|
|
Args:
|
|||
|
|
keywords: 关键词列表
|
|||
|
|
brand: 品牌名称
|
|||
|
|
days: 预测未来多少天
|
|||
|
|
|
|||
|
|
Returns:
|
|||
|
|
每个关键词的趋势预测：
|
|||
|
|
- trend: 趋势方向（上升/下降/稳定）
|
|||
|
|
- trend_strength: 趋势强度（0-1）
|
|||
|
|
- predicted_mention_rate: 预测提及率
|
|||
|
|
- confidence: 预测置信度（0-1）
|
|||
|
|
"""
|
|||
|
|
verify_df = self.storage.get_verify_results(brand=brand, include_timestamp=True)
|
|||
|
|
|
|||
|
|
if verify_df.empty or "验证时间" not in verify_df.columns:
|
|||
|
|
# 如果没有时间戳数据，返回默认值
|
|||
|
|
return {
|
|||
|
|
kw: {
|
|||
|
|
"trend": "未知",
|
|||
|
|
"trend_strength": 0.0,
|
|||
|
|
"predicted_mention_rate": 0.0,
|
|||
|
|
"confidence": 0.0
|
|||
|
|
}
|
|||
|
|
for kw in keywords
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
trend_analysis = {}
|
|||
|
|
|
|||
|
|
for keyword in keywords:
|
|||
|
|
keyword_data = verify_df[verify_df["问题"] == keyword].copy()
|
|||
|
|
|
|||
|
|
if keyword_data.empty:
|
|||
|
|
trend_analysis[keyword] = {
|
|||
|
|
"trend": "未知",
|
|||
|
|
"trend_strength": 0.0,
|
|||
|
|
"predicted_mention_rate": 0.0,
|
|||
|
|
"confidence": 0.0
|
|||
|
|
}
|
|||
|
|
continue
|
|||
|
|
|
|||
|
|
# 转换时间列
|
|||
|
|
keyword_data["验证时间"] = pd.to_datetime(keyword_data["验证时间"])
|
|||
|
|
keyword_data = keyword_data.sort_values("验证时间")
|
|||
|
|
|
|||
|
|
# 按日期分组，计算每天的提及率
|
|||
|
|
keyword_data["日期"] = keyword_data["验证时间"].dt.date
|
|||
|
|
daily_stats = keyword_data.groupby("日期").agg({
|
|||
|
|
"提及次数": "mean",
|
|||
|
|
# 提及率 = 当天该品牌出现次数 / 当天总记录数，防御性处理空分组
|
|||
|
|
"品牌": lambda x: (x == brand).sum() / len(x) if len(x) > 0 else 0.0
|
|||
|
|
}).reset_index()
|
|||
|
|
daily_stats.columns = ["日期", "平均提及次数", "提及率"]
|
|||
|
|
|
|||
|
|
if len(daily_stats) < 2:
|
|||
|
|
# 数据点太少，无法预测
|
|||
|
|
trend_analysis[keyword] = {
|
|||
|
|
"trend": "数据不足",
|
|||
|
|
"trend_strength": 0.0,
|
|||
|
|
"predicted_mention_rate": daily_stats["提及率"].iloc[-1] if len(daily_stats) > 0 else 0.0,
|
|||
|
|
"confidence": 0.0
|
|||
|
|
}
|
|||
|
|
continue
|
|||
|
|
|
|||
|
|
# 计算趋势（简单线性回归）
|
|||
|
|
x = list(range(len(daily_stats)))
|
|||
|
|
y = daily_stats["提及率"].values
|
|||
|
|
|
|||
|
|
# 计算斜率（趋势方向）
|
|||
|
|
n = len(x)
|
|||
|
|
x_mean = sum(x) / n
|
|||
|
|
y_mean = sum(y) / n
|
|||
|
|
|
|||
|
|
numerator = sum((x[i] - x_mean) * (y[i] - y_mean) for i in range(n))
|
|||
|
|
denominator = sum((x[i] - x_mean) ** 2 for i in range(n))
|
|||
|
|
|
|||
|
|
if denominator == 0:
|
|||
|
|
slope = 0
|
|||
|
|
else:
|
|||
|
|
slope = numerator / denominator
|
|||
|
|
|
|||
|
|
# 判断趋势
|
|||
|
|
if slope > 0.01:
|
|||
|
|
trend = "上升"
|
|||
|
|
trend_strength = min(abs(slope) * 10, 1.0)
|
|||
|
|
elif slope < -0.01:
|
|||
|
|
trend = "下降"
|
|||
|
|
trend_strength = min(abs(slope) * 10, 1.0)
|
|||
|
|
else:
|
|||
|
|
trend = "稳定"
|
|||
|
|
trend_strength = 0.0
|
|||
|
|
|
|||
|
|
# 预测未来提及率（简单线性外推）
|
|||
|
|
current_rate = y[-1]
|
|||
|
|
predicted_rate = current_rate + slope * (days / len(daily_stats))
|
|||
|
|
predicted_rate = max(0.0, min(1.0, predicted_rate)) # 限制在 0-1 之间
|
|||
|
|
|
|||
|
|
# 计算置信度（基于数据点数量）
|
|||
|
|
confidence = min(len(daily_stats) / 10, 1.0) # 10个数据点达到最高置信度
|
|||
|
|
|
|||
|
|
trend_analysis[keyword] = {
|
|||
|
|
"trend": trend,
|
|||
|
|
"trend_strength": trend_strength,
|
|||
|
|
"predicted_mention_rate": predicted_rate,
|
|||
|
|
"confidence": confidence,
|
|||
|
|
"current_rate": current_rate,
|
|||
|
|
"data_points": len(daily_stats)
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return trend_analysis
|
|||
|
|
|
|||
|
|
def calculate_value_matrix(
|
|||
|
|
self,
|
|||
|
|
keywords: List[str],
|
|||
|
|
competition_data: Dict[str, Dict[str, Any]],
|
|||
|
|
estimated_values: Optional[Dict[str, float]] = None
|
|||
|
|
) -> Dict[str, Dict[str, Any]]:
|
|||
|
|
"""
|
|||
|
|
计算关键词价值矩阵
|
|||
|
|
|
|||
|
|
Args:
|
|||
|
|
keywords: 关键词列表
|
|||
|
|
competition_data: 竞争度分析结果
|
|||
|
|
estimated_values: 预估价值字典（可选）
|
|||
|
|
|
|||
|
|
Returns:
|
|||
|
|
每个关键词的价值矩阵分析：
|
|||
|
|
- value_score: 价值分数（0-10）
|
|||
|
|
- competition_score: 竞争分数（0-10，越高竞争越激烈）
|
|||
|
|
- matrix_position: 矩阵位置（高价值低竞争/高价值高竞争/低价值低竞争/低价值高竞争）
|
|||
|
|
- recommendation: 推荐建议
|
|||
|
|
"""
|
|||
|
|
value_matrix = {}
|
|||
|
|
|
|||
|
|
for keyword in keywords:
|
|||
|
|
comp_data = competition_data.get(keyword, {})
|
|||
|
|
|
|||
|
|
# 计算价值分数（基于预估价值和提及率）
|
|||
|
|
if estimated_values and keyword in estimated_values:
|
|||
|
|
value_score = estimated_values[keyword]
|
|||
|
|
else:
|
|||
|
|
# 如果没有预估价值，基于提及率估算
|
|||
|
|
mention_rate = comp_data.get("mention_rate", 0.0)
|
|||
|
|
value_score = mention_rate * 10 # 转换为 0-10 分
|
|||
|
|
|
|||
|
|
# 计算竞争分数（基于竞争级别）
|
|||
|
|
competition_level = comp_data.get("competition_level", "未知")
|
|||
|
|
if competition_level == "高":
|
|||
|
|
competition_score = 8
|
|||
|
|
elif competition_level == "中":
|
|||
|
|
competition_score = 5
|
|||
|
|
elif competition_level == "低":
|
|||
|
|
competition_score = 2
|
|||
|
|
else:
|
|||
|
|
competition_score = 5 # 默认中等
|
|||
|
|
|
|||
|
|
# 判断矩阵位置
|
|||
|
|
if value_score >= 6 and competition_score <= 4:
|
|||
|
|
matrix_position = "高价值低竞争"
|
|||
|
|
recommendation = "强烈推荐：高价值且竞争低，优先投入"
|
|||
|
|
elif value_score >= 6 and competition_score > 4:
|
|||
|
|
matrix_position = "高价值高竞争"
|
|||
|
|
recommendation = "谨慎投入：价值高但竞争激烈，需要持续优化"
|
|||
|
|
elif value_score < 6 and competition_score <= 4:
|
|||
|
|
matrix_position = "低价值低竞争"
|
|||
|
|
recommendation = "可考虑：价值一般但竞争低，适合长尾策略"
|
|||
|
|
else:
|
|||
|
|
matrix_position = "低价值高竞争"
|
|||
|
|
recommendation = "不推荐：价值低且竞争激烈，避免投入"
|
|||
|
|
|
|||
|
|
value_matrix[keyword] = {
|
|||
|
|
"value_score": round(value_score, 2),
|
|||
|
|
"competition_score": competition_score,
|
|||
|
|
"matrix_position": matrix_position,
|
|||
|
|
"recommendation": recommendation
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
return value_matrix
|
|||
|
|
|
|||
|
|
def mine_longtail_keywords(
|
|||
|
|
self,
|
|||
|
|
base_keywords: List[str],
|
|||
|
|
brand: str,
|
|||
|
|
advantages: str,
|
|||
|
|
num_longtail: int = 15,
|
|||
|
|
llm_chain=None
|
|||
|
|
) -> List[Dict[str, Any]]:
|
|||
|
|
"""
|
|||
|
|
挖掘长尾关键词（关键词组合）
|
|||
|
|
|
|||
|
|
Args:
|
|||
|
|
base_keywords: 基础关键词列表
|
|||
|
|
brand: 品牌名称
|
|||
|
|
advantages: 品牌优势
|
|||
|
|
num_longtail: 需要生成的长尾词数量
|
|||
|
|
llm_chain: LLM 调用链（可选）
|
|||
|
|
|
|||
|
|
Returns:
|
|||
|
|
长尾关键词列表
|
|||
|
|
"""
|
|||
|
|
if not llm_chain or not base_keywords:
|
|||
|
|
return []
|
|||
|
|
|
|||
|
|
# 选择前10个基础关键词
|
|||
|
|
selected_base = base_keywords[:10]
|
|||
|
|
|
|||
|
|
prompt = f"""你是长尾关键词挖掘专家，基于基础关键词生成长尾关键词。
|
|||
|
|
|
|||
|
|
【品牌信息】
|
|||
|
|
- 品牌：{brand}
|
|||
|
|
- 核心优势：{advantages}
|
|||
|
|
|
|||
|
|
【基础关键词】
|
|||
|
|
{json.dumps(selected_base, ensure_ascii=False, indent=2)}
|
|||
|
|
|
|||
|
|
【任务】
|
|||
|
|
基于这些基础关键词，生成 {num_longtail} 个长尾关键词，要求：
|
|||
|
|
1. 在基础关键词上添加修饰词（如：最好的、免费的、2025年、如何等）
|
|||
|
|
2. 结合用户搜索意图（对比、评测、使用、购买等）
|
|||
|
|
3. 自然、口语化，12-28字
|
|||
|
|
4. 与品牌和优势相关
|
|||
|
|
|
|||
|
|
【输出格式】
|
|||
|
|
JSON 数组，每个长尾词包含：
|
|||
|
|
- keyword: 长尾关键词文本
|
|||
|
|
- base_keyword: 对应的基础关键词
|
|||
|
|
- modifier: 添加的修饰词
|
|||
|
|
- intent: 搜索意图
|
|||
|
|
|
|||
|
|
【示例】
|
|||
|
|
[
|
|||
|
|
{{
|
|||
|
|
"keyword": "2025年最好的{selected_base[0]}",
|
|||
|
|
"base_keyword": "{selected_base[0]}",
|
|||
|
|
"modifier": "2025年最好的",
|
|||
|
|
"intent": "对比选择"
|
|||
|
|
}}
|
|||
|
|
]
|
|||
|
|
|
|||
|
|
【开始输出 JSON 数组】
|
|||
|
|
"""
|
|||
|
|
|
|||
|
|
try:
|
|||
|
|
result = llm_chain.invoke({"input": prompt})
|
|||
|
|
|
|||
|
|
# 解析 JSON
|
|||
|
|
if isinstance(result, str):
|
|||
|
|
import re
|
|||
|
|
json_match = re.search(r'\[[\s\S]*\]', result)
|
|||
|
|
if json_match:
|
|||
|
|
longtail_keywords = json.loads(json_match.group(0))
|
|||
|
|
else:
|
|||
|
|
longtail_keywords = []
|
|||
|
|
else:
|
|||
|
|
longtail_keywords = result if isinstance(result, list) else []
|
|||
|
|
|
|||
|
|
# 验证和清理
|
|||
|
|
cleaned = []
|
|||
|
|
for kw in longtail_keywords:
|
|||
|
|
if isinstance(kw, dict) and "keyword" in kw:
|
|||
|
|
cleaned.append({
|
|||
|
|
"keyword": kw.get("keyword", ""),
|
|||
|
|
"base_keyword": kw.get("base_keyword", ""),
|
|||
|
|
"modifier": kw.get("modifier", ""),
|
|||
|
|
"intent": kw.get("intent", "未知")
|
|||
|
|
})
|
|||
|
|
|
|||
|
|
return cleaned[:num_longtail]
|
|||
|
|
|
|||
|
|
except Exception as e:
|
|||
|
|
print(f"长尾词挖掘失败: {e}")
|
|||
|
|
return []
|
|||
|
|
|
|||
|
|
def recommend_keywords(
|
|||
|
|
self,
|
|||
|
|
keywords: List[str],
|
|||
|
|
value_matrix: Dict[str, Dict[str, Any]],
|
|||
|
|
competition_data: Dict[str, Dict[str, Any]],
|
|||
|
|
trend_data: Optional[Dict[str, Dict[str, Any]]] = None,
|
|||
|
|
top_n: int = 10
|
|||
|
|
) -> List[Dict[str, Any]]:
|
|||
|
|
"""
|
|||
|
|
智能推荐最优关键词
|
|||
|
|
|
|||
|
|
Args:
|
|||
|
|
keywords: 关键词列表
|
|||
|
|
value_matrix: 价值矩阵分析结果
|
|||
|
|
competition_data: 竞争度分析结果
|
|||
|
|
trend_data: 趋势预测数据（可选）
|
|||
|
|
top_n: 返回前 N 个推荐
|
|||
|
|
|
|||
|
|
Returns:
|
|||
|
|
推荐的关键词列表，按推荐度排序
|
|||
|
|
"""
|
|||
|
|
recommendations = []
|
|||
|
|
|
|||
|
|
for keyword in keywords:
|
|||
|
|
value_info = value_matrix.get(keyword, {})
|
|||
|
|
comp_info = competition_data.get(keyword, {})
|
|||
|
|
trend_info = trend_data.get(keyword, {}) if trend_data else {}
|
|||
|
|
|
|||
|
|
# 计算推荐分数
|
|||
|
|
value_score = value_info.get("value_score", 0)
|
|||
|
|
competition_score = value_info.get("competition_score", 5)
|
|||
|
|
trend_strength = trend_info.get("trend_strength", 0) if trend_info else 0
|
|||
|
|
trend_direction = trend_info.get("trend", "稳定") if trend_info else "稳定"
|
|||
|
|
|
|||
|
|
# 推荐分数 = 价值分数 - 竞争分数 + 趋势加分
|
|||
|
|
# 趋势加分：上升趋势 +2，下降趋势 -1
|
|||
|
|
trend_bonus = 2 if trend_direction == "上升" else (-1 if trend_direction == "下降" else 0)
|
|||
|
|
trend_bonus = trend_bonus * trend_strength # 根据趋势强度调整
|
|||
|
|
|
|||
|
|
recommendation_score = value_score - (competition_score / 2) + trend_bonus
|
|||
|
|
|
|||
|
|
recommendations.append({
|
|||
|
|
"keyword": keyword,
|
|||
|
|
"recommendation_score": round(recommendation_score, 2),
|
|||
|
|
"value_score": value_score,
|
|||
|
|
"competition_score": competition_score,
|
|||
|
|
"trend": trend_direction,
|
|||
|
|
"matrix_position": value_info.get("matrix_position", "未知"),
|
|||
|
|
"recommendation": value_info.get("recommendation", "")
|
|||
|
|
})
|
|||
|
|
|
|||
|
|
# 按推荐分数排序
|
|||
|
|
recommendations.sort(key=lambda x: x["recommendation_score"], reverse=True)
|
|||
|
|
|
|||
|
|
return recommendations[:top_n]
|