@@ -53,13 +53,13 @@ Unlike traditional memory tools that treat knowledge as static data to be retrie
- Domain jargon, colloquial expressions, and context-dependent references are not accurately encoded, leading to semantic drift in memory interpretation
- Cross-language memory associations fail in multilingual or dialect-rich scenarios
-
+
---
## Core Features
-
+
### Memory Extraction Engine
@@ -121,7 +121,7 @@ Unified service architecture exposing two API surfaces:
## Architecture
-
+
**Celery Three-Queue Async Architecture:**
@@ -139,15 +139,15 @@ Evaluation metrics include F1 score (F1), BLEU-1 (B1), and LLM-as-a-Judge score
MemoryBear consistently outperforms competing systems including Mem0, Zep, and LangMem across all four task categories:
-
+
**Vector version (non-graph)**: Achieves substantially improved retrieval efficiency while maintaining high accuracy. Overall accuracy surpasses the best existing full-text retrieval methods (72.90 ± 0.19%), while maintaining low latency at both p50 and p95 for Search Latency and Total Latency.
-
+
**Graph version**: Integrating the knowledge graph architecture pushes overall accuracy to a new benchmark (**75.00 ± 0.20%**), delivering performance metrics that significantly surpass all other methods.
-
+
---
@@ -229,7 +229,7 @@ npm install && npm run dev
git clone https://github.com/SuanmoSuanyangTechnology/MemoryBear.git
```
-
+
### 3. Backend API Service
@@ -260,19 +260,19 @@ Download [Docker Desktop](https://www.docker.com/products/docker-desktop/) and p
**PostgreSQL** — search → select → pull
-
+
-
+
-
+
**Neo4j** — pull the same way. When creating the container, map two required ports and set an initial password:
- `7474`: Neo4j Browser
- `7687`: Bolt protocol
-
+
-
+
**Redis** — same steps as above.
@@ -343,9 +343,9 @@ Apply all migrations to create the full schema:
alembic upgrade head
```
-
+
-
+
#### 3.5 Start the API Service
@@ -355,7 +355,7 @@ uv run -m app.main
Access API documentation at http://localhost:8000/docs
-
+
#### 3.6 Start Celery Workers (Optional, for async tasks)
@@ -401,9 +401,9 @@ proxy: {
npm run dev
```
-
+
-
+
### 5. Initialize the System
diff --git a/README_CN.md b/README_CN.md
index 4fbe88a7..e2c984ed 100644
--- a/README_CN.md
+++ b/README_CN.md
@@ -1,4 +1,4 @@
-
+
@@ -53,13 +53,13 @@ MemoryBear 是红熊 AI 自主研发的新一代 AI 记忆系统，核心突破
- 行业术语、口语化表达、上下文指代未被准确编码，导致模型对记忆内容的语义解析失真
- 多语言混用场景中，跨语种记忆关联失效
-
+
---
## 核心特性
-
+
### 记忆萃取引擎
@@ -120,7 +120,7 @@ MemoryBear 是红熊 AI 自主研发的新一代 AI 记忆系统，核心突破
## 架构总览
-
+
**Celery 三队列异步架构：**
@@ -138,15 +138,15 @@ MemoryBear 是红熊 AI 自主研发的新一代 AI 记忆系统，核心突破
MemoryBear 在四大任务类型的核心指标中，均优于行业内竞争对手 Mem0、Zep、LangMem 等现有方法：
-
+
**向量版本（非图谱）**：在保持高准确性的同时极大优化了检索效率，总体准确性明显高于现有最高全文检索方法（72.90 ± 0.19%），且在 Search Latency 和 Total Latency 的 p50/p95 上保持较低水平。
-
+
**图谱版本**：通过集成知识图谱架构，将总体准确性推至新高度（**75.00 ± 0.20%**），在保持准确性的同时整体指标显著优于所有其他方法。
-
+
---
@@ -228,7 +228,7 @@ npm install && npm run dev
git clone https://github.com/SuanmoSuanyangTechnology/MemoryBear.git
```
-
+
### 三、后端 API 服务启动
@@ -261,13 +261,13 @@ source .venv/bin/activate
拉取镜像：search → select → pull
-
+
创建容器：
-
+
-
+
**Neo4j**
@@ -275,9 +275,9 @@ source .venv/bin/activate
- `7474`：Neo4j Browser
- `7687`：Bolt 协议
-
+
-
+
**Redis**：同上步骤拉取并创建容器。
@@ -348,9 +348,9 @@ sqlalchemy.url = postgresql://用户名:密码@数据库地址:端口/数据库
alembic upgrade head
```
-
+
-
+
#### 3.5 启动 API 服务
@@ -360,7 +360,7 @@ uv run -m app.main
访问 API 文档：http://localhost:8000/docs
-
+
#### 3.6 启动 Celery Worker（可选，用于异步任务）
@@ -406,9 +406,9 @@ proxy: {
npm run dev
```
-
+
-
+
### 五、初始化系统