feat(memory): 添加用户画像功能支持

- 新增 `enable_user_profile` 配置项，支持用户画像模式 - 实现 `UserMemory` 类集成，自动提取用户信息 - 更新文档说明用户画像功能及配置要求
2026-07-21 22:53:56 +08:00 · 2026-01-08 14:43:48 +08:00
parent dadf05ab4e
commit de7aedab96
3 changed files with 189 additions and 45 deletions
@@ -17,6 +17,8 @@
 ## 功能特性

 - **本地总结**：通过 LLM 在本地进行记忆总结和提取
+- **用户画像**：通过 `UserMemory` 自动提取用户信息（姓名、职业、兴趣等），持续更新用户画像
+- **智能遗忘**：基于艾宾浩斯遗忘曲线，自动"遗忘"过时噪声信息
 - **多种存储后端**：支持 OceanBase（推荐，最佳性能）、SeekDB（推荐，AI应用存储一体）、PostgreSQL、SQLite（轻量备选）
 - **多种 LLM 支持**：通义千问、智谱（glm-4-flash 免费）、OpenAI 等
 - **智能检索**：基于向量搜索的语义检索能力
@@ -44,6 +46,8 @@ selected_module:
 Memory:
  powermem:
    type: powermem
+    # 是否启用用户画像功能（需要OceanBase）
+    enable_user_profile: false
    # 数据库提供者: oceanbase(推荐,最佳性能), seekdb, postgres, sqlite(轻量备选)
    database_provider: sqlite  # 资源充足时建议使用 oceanbase 或 seekdb
    # LLM提供者: qwen(默认), openai, 等
@@ -62,6 +66,7 @@ Memory:

 | 参数 | 说明 | 默认值 | 可选值 |
 |------|------|--------|--------|
+| `enable_user_profile` | 启用用户画像模式 | `false` | `true`(需OceanBase), `false` |
 | `database_provider` | 存储后端类型 | `sqlite` | `oceanbase`(推荐), `seekdb`, `postgres`, `sqlite`(轻量) |
 | `llm_provider` | LLM 提供商 | `qwen` | `qwen`, `zhipu`(免费), `openai`, 等 |
 | `embedding_provider` | 嵌入模型提供商 | `qwen` | `qwen`, `zhipu`, `openai`, 等 |
@@ -72,6 +77,15 @@ Memory:
 | `embedding_model` | 嵌入模型名称 | - | 根据提供商选择 |
 | `embedding_base_url` | 嵌入模型 API 地址（可选） | - | - |

+### 记忆模式说明
+
+PowerMem 支持两种记忆模式：
+
+| 模式 | 配置 | 功能 | 存储要求 |
+|------|------|------|----------|
+| **普通记忆** | `enable_user_profile: false` | 对话记忆存储与检索 | 支持所有数据库 |
+| **用户画像** | `enable_user_profile: true` | 记忆 + 自动提取用户画像 | 仅支持 OceanBase |
+
 ### 使用通义千问（推荐）

 1. 访问 [阿里云百炼平台](https://bailian.console.aliyun.com/) 注册账号
@@ -198,6 +212,54 @@ PowerMem 会自动使用设备 ID（`device_id`）作为 `user_id` 进行记忆
 - 不同设备之间的记忆完全隔离
 - 同一设备的多次对话可以共享记忆上下文

+## 用户画像（UserMemory）
+
+PowerMem 提供 `UserMemory` 类，可自动从对话中提取用户画像信息。
+
+### 启用用户画像
+
+在配置中设置 `enable_user_profile: true` 即可启用：
+
+```yaml
+Memory:
+  powermem:
+    type: powermem
+    enable_user_profile: true  # 启用用户画像
+    database_provider: oceanbase  # 必须使用 OceanBase
+    llm_provider: qwen
+    embedding_provider: qwen
+    llm_api_key: sk-xxxxxxxxxxxxxxxx
+    llm_model: qwen-plus
+    embedding_api_key: sk-xxxxxxxxxxxxxxxx
+    embedding_model: text-embedding-v3
+    # OceanBase 数据库连接配置
+    vector_store:
+      provider: oceanbase
+      config:
+        host: 127.0.0.1
+        port: 2881
+        user: root@test
+        password: your_password
+        database: powermem
+```
+
+### 用户画像能力
+
+| 能力 | 说明 |
+|------|------|
+| **信息提取** | 自动从对话中提取姓名、年龄、职业、兴趣等 |
+| **持续更新** | 随着对话进行，不断完善用户画像 |
+| **画像检索** | 将用户画像与记忆搜索结合，提升检索相关性 |
+| **智能遗忘** | 基于艾宾浩斯遗忘曲线，淡化过时信息 |
+
+### 工作原理
+
+启用用户画像后，小智在查询记忆时会自动返回：
+1. **用户画像**：用户的基本信息、兴趣爱好等
+2. **相关记忆**：与当前对话相关的历史记忆
+
+> ⚠️ **注意**：`UserMemory` 功能需要 OceanBase 作为存储后端，其他数据库暂不支持。
+
 ## 与其他记忆组件的对比

 | 特性 | PowerMem | mem0ai | mem_local_short |
@@ -206,6 +268,8 @@ PowerMem 会自动使用设备 ID（`device_id`）作为 `user_id` 进行记忆
 | 存储位置 | 本地/云端DB | 云端 | 本地YAML |
 | 费用 | 取决于LLM和DB | 1000次/月免费 | 完全免费 |
 | 智能检索 | ✅ 向量搜索 | ✅ 向量搜索 | ❌ 全量返回 |
+| 用户画像 | ✅ UserMemory | ❌ | ❌ |
+| 智能遗忘 | ✅ 遗忘曲线 | ❌ | ❌ |
 | 私有部署 | ✅ 支持 | ❌ 仅云端 | ✅ 支持 |
 | 数据库支持 | OceanBase(推荐)/SeekDB/PostgreSQL/SQLite | - | YAML 文件 |

@@ -239,6 +303,9 @@ source .venv/bin/activate

 # 测试 PowerMem 导入
 python -c "from powermem import AsyncMemory; print('PowerMem 导入成功')"
+
+# 测试 UserMemory 导入（用户画像功能）
+python -c "from powermem import UserMemory; print('UserMemory 导入成功')"
 ```

 ## 更多资源
@@ -284,8 +284,13 @@ Memory:
    # 官网: https://www.powermem.ai/
    # 使用示例: https://github.com/oceanbase/powermem/tree/main/examples
    type: powermem
+    # 是否启用用户画像功能（需要OceanBase作为存储后端）
+    # - false: 使用普通记忆模式(AsyncMemory)，支持所有数据库
+    # - true: 使用用户画像模式(UserMemory)，自动提取用户信息，仅支持OceanBase
+    enable_user_profile: false
    # 数据库提供者: oceanbase(推荐,最佳性能), seekdb(推荐,AI应用存储一体), postgres, sqlite(轻量备选)
    # 资源充足时建议使用 oceanbase 或 seekdb
+    # 注意：用户画像功能(enable_user_profile: true)仅支持oceanbase
    database_provider: sqlite
    # LLM提供者: qwen(默认), zhipu(免费glm-4-flash), openai, 等
    llm_provider: qwen
@@ -7,13 +7,13 @@
       PowerMem is an open-source agent memory component from OceanBase
       GitHub: https://github.com/oceanbase/powermem
       Website: https://www.powermem.ai/
+@Author: wayyoungboy
 """

 import traceback
 from typing import Optional, Dict, Any

 from ..base import MemoryProviderBase, logger
-from powermem import AsyncMemory

 TAG = __name__

@@ -27,19 +27,37 @@ class MemoryProvider(MemoryProviderBase):
    
    Supports multiple storage backends (sqlite, oceanbase, postgres),
    LLM providers (qwen, openai, etc.) and embedding providers.
+    
+    Config options:
+        - enable_user_profile: bool - Enable UserMemory for user profiling (requires OceanBase)
+        - database_provider: str - Storage backend (sqlite, oceanbase, postgres)
+        - llm_provider: str - LLM provider (qwen, openai, etc.)
+        - embedding_provider: str - Embedding provider (qwen, openai, etc.)
    """

    def __init__(self, config: Dict[str, Any], summary_memory: Optional[str] = None):
        super().__init__(config)
        self.use_powermem = False
        self.memory_client = None
+        self.enable_user_profile = False
        
-        try:            
+        try:
+            # Check if user profile mode is enabled
+            self.enable_user_profile = config.get("enable_user_profile", False)
+            
            # Get configuration parameters
            database_provider = config.get("database_provider", "sqlite")
            llm_provider = config.get("llm_provider", "qwen")
            embedding_provider = config.get("embedding_provider", "qwen")
            
+            # UserMemory requires OceanBase
+            if self.enable_user_profile and database_provider not in ["oceanbase"]:
+                logger.bind(tag=TAG).warning(
+                    f"UserMemory requires OceanBase as storage backend, but got {database_provider}. "
+                    "Falling back to AsyncMemory mode."
+                )
+                self.enable_user_profile = False
+            
            # Build powermem configuration dict
            # PowerMem supports two config styles:
            # 1. powermem style: database, llm, embedding
@@ -89,13 +107,21 @@ class MemoryProvider(MemoryProviderBase):
                    "config": embedder_config
                }
            
-            # Initialize AsyncMemory client
-            self.memory_client = AsyncMemory(config=powermem_config)
+            # Initialize memory client based on mode
+            if self.enable_user_profile:
+                from powermem import UserMemory
+                self.memory_client = UserMemory(config=powermem_config)
+                memory_mode = "UserMemory (用户画像模式)"
+            else:
+                from powermem import AsyncMemory
+                self.memory_client = AsyncMemory(config=powermem_config)
+                memory_mode = "AsyncMemory (普通记忆模式)"
+            
            self.use_powermem = True
            
            logger.bind(tag=TAG).info(
-                f"PowerMem initialized successfully with database={database_provider}, "
-                f"llm={llm_provider}, embedding={embedding_provider}"
+                f"PowerMem initialized successfully: mode={memory_mode}, "
+                f"database={database_provider}, llm={llm_provider}, embedding={embedding_provider}"
            )
            
        except ImportError as e:
@@ -167,6 +193,14 @@ class MemoryProvider(MemoryProviderBase):
                logger.bind(tag=TAG).debug("No role_id set, returning empty memory")
                return ""

+            result_parts = []
+            
+            # If user profile mode is enabled, include user profile in results
+            if self.enable_user_profile:
+                profile = await self.get_user_profile()
+                if profile:
+                    result_parts.append(f"【用户画像】\n{profile}")
+
            # Search memories using PowerMem SDK
            results = await self.memory_client.search(
                query=query,
@@ -174,54 +208,92 @@ class MemoryProvider(MemoryProviderBase):
                limit=30
            )
                    
-            if not results or "results" not in results:
-                logger.bind(tag=TAG).debug("No memory results found")
-                return ""
-
-            # Format each memory entry with its update time
-            memories = []
-            for entry in results.get("results", []):
-                # Get timestamp from updated_at or created_at
-                timestamp = ""
-                if "updated_at" in entry and entry["updated_at"]:
-                    timestamp = str(entry["updated_at"])
-                elif "created_at" in entry and entry["created_at"]:
-                    timestamp = str(entry["created_at"])
-                    
-                if timestamp:
-                    try:
-                        # Parse and reformat the timestamp (remove milliseconds if present)
-                        if "." in timestamp:
-                            dt = timestamp.split(".")[0]
-                        else:
-                            dt = timestamp
-                        formatted_time = dt.replace("T", " ")
-                    except Exception:
-                        formatted_time = timestamp
-                else:
-                    formatted_time = ""
-                    
-                memory = entry.get("memory", "") or entry.get("content", "")
-                if memory:
-                    if formatted_time:
-                        # Store tuple of (timestamp, formatted_string) for sorting
-                        memories.append((timestamp, f"[{formatted_time}] {memory}"))
+            if results and "results" in results:
+                # Format each memory entry with its update time
+                memories = []
+                for entry in results.get("results", []):
+                    # Get timestamp from updated_at or created_at
+                    timestamp = ""
+                    if "updated_at" in entry and entry["updated_at"]:
+                        timestamp = str(entry["updated_at"])
+                    elif "created_at" in entry and entry["created_at"]:
+                        timestamp = str(entry["created_at"])
+                        
+                    if timestamp:
+                        try:
+                            # Parse and reformat the timestamp (remove milliseconds if present)
+                            if "." in timestamp:
+                                dt = timestamp.split(".")[0]
+                            else:
+                                dt = timestamp
+                            formatted_time = dt.replace("T", " ")
+                        except Exception:
+                            formatted_time = timestamp
                    else:
-                        memories.append(("", memory))
+                        formatted_time = ""
+                        
+                    memory = entry.get("memory", "") or entry.get("content", "")
+                    if memory:
+                        if formatted_time:
+                            # Store tuple of (timestamp, formatted_string) for sorting
+                            memories.append((timestamp, f"[{formatted_time}] {memory}"))
+                        else:
+                            memories.append(("", memory))

-            # Sort by timestamp in descending order (newest first)
-            memories.sort(key=lambda x: x[0], reverse=True)
+                # Sort by timestamp in descending order (newest first)
+                memories.sort(key=lambda x: x[0], reverse=True)

-            # Extract only the formatted strings
-            memories_str = "\n".join(f"- {memory[1]}" for memory in memories)
-            logger.bind(tag=TAG).debug(f"Query results: {memories_str}")
-            return memories_str
+                # Extract only the formatted strings
+                if memories:
+                    memories_str = "\n".join(f"- {memory[1]}" for memory in memories)
+                    result_parts.append(f"【相关记忆】\n{memories_str}")
+
+            final_result = "\n\n".join(result_parts)
+            logger.bind(tag=TAG).debug(f"Query results: {final_result}")
+            return final_result
            
        except Exception as e:
            logger.bind(tag=TAG).error(f"Error querying memory: {str(e)}")
            logger.bind(tag=TAG).debug(f"Detailed error: {traceback.format_exc()}")
            return ""

+    async def get_user_profile(self) -> str:
+        """
+        Get user profile from PowerMem (only available in UserMemory mode).
+        
+        Returns:
+            Formatted user profile string or empty string if not available
+        """
+        if not self.use_powermem or self.memory_client is None:
+            return ""
+            
+        if not self.enable_user_profile:
+            logger.bind(tag=TAG).debug("User profile mode is not enabled")
+            return ""
+            
+        try:
+            if not getattr(self, "role_id", None):
+                return ""
+                
+            # Get user profile using UserMemory SDK
+            profile = await self.memory_client.get_profile(user_id=self.role_id)
+            
+            if not profile:
+                return ""
+            
+            # Format profile as readable string
+            profile_parts = []
+            for key, value in profile.items():
+                if value:
+                    profile_parts.append(f"- {key}: {value}")
+            
+            return "\n".join(profile_parts)
+            
+        except Exception as e:
+            logger.bind(tag=TAG).error(f"Error getting user profile: {str(e)}")
+            logger.bind(tag=TAG).debug(f"Detailed error: {traceback.format_exc()}")
+            return ""
+

 # Register the memory provider instance
 powermem = MemoryProvider({})