Skip to content

Commit 2911f9f

Browse files
andylau-55andylau-55
committed
add website #andy
1 parent aa43b57 commit 2911f9f

File tree

163 files changed

+7721
-607
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

163 files changed

+7721
-607
lines changed

‎.idea/workspace.xml

Lines changed: 17 additions & 33 deletions
Some generated files are not rendered by default. Learn more about customizing how changed files appear on GitHub.

‎website/docs/ch/_category_.json

Lines changed: 7 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,7 @@
1+
{
2+
"label": "中文文档",
3+
"position": 2,
4+
"link": {
5+
"type": "generated-index"
6+
}
7+
}

‎website/docs/ch/index.md

Lines changed: 81 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,81 @@
1+
# 用户手册
2+
3+
4+
## [快速开始](快速开始.md)
5+
6+
## 用户向导
7+
8+
- [用户权限管理](用户向导/用户权限管理.md)
9+
- [知识库配置](用户向导/知识库配置.md)
10+
11+
### 模型服务配置
12+
13+
- [表示(embedding)模型](用户向导/模型服务配置/表示(embedding)模型.md)
14+
- [生成(chat)模型](用户向导/模型服务配置/生成(chat)模型.md)
15+
16+
### 知识建模
17+
18+
- [声明式schema](用户向导/知识建模/声明式schema.md)
19+
- [建模最佳实践](用户向导/知识建模/建模最佳实践.md)
20+
- [概念关系语义分类](用户向导/知识建模/概念关系语义分类.md)
21+
- [可视化查看](用户向导/知识建模/可视化查看.md)
22+
23+
### 知识推理语法
24+
25+
- [专家规则（DSL）语法](用户向导/知识推理语法/专家规则（DSL）语法.md)
26+
- [物理引擎适配RDG](用户向导/知识推理语法/物理引擎适配RDG.md)
27+
28+
### 自定义扩展
29+
30+
- [自定义schema](用户向导/自定义扩展/自定义schema.md)
31+
- [自定义prompt](用户向导/自定义扩展/自定义prompt.md)
32+
- [自定义代码](用户向导/自定义扩展/自定义代码.md)
33+
- [领域知识挂载](用户向导/自定义扩展/领域知识挂载.md)
34+
- [知识探查](用户向导/知识探查.md)
35+
- [命令行工具](用户向导/命令行工具.md)
36+
- [源码编译&部署](用户向导/源码编译&部署.md)
37+
- [v0.5如何升级到v0.6](用户向导/v0.5如何升级到v0.6.md)
38+
39+
## 内置案例
40+
41+
42+
### 垂域数据集
43+
44+
- [企业供应链](内置案例/垂域数据集/企业供应链.md)
45+
- [黑产挖掘](内置案例/垂域数据集/黑产挖掘.md)
46+
- [医疗图谱](内置案例/垂域数据集/医疗图谱.md)
47+
48+
### 公开数据集
49+
50+
- [benchmark](内置案例/公开数据集/benchmark.md)
51+
- [hotpotqa](内置案例/公开数据集/hotpotqa.md)
52+
- [musique](内置案例/公开数据集/musique.md)
53+
- [twowiki](内置案例/公开数据集/twowiki.md)
54+
55+
## 系统集成
56+
57+
- [HTTPAPI Reference](系统集成/HTTPAPIReference.md)
58+
59+
## 设计理念&代码架构
60+
61+
62+
### 设计理念
63+
64+
- [KAG技术报告](设计理念&代码架构/设计理念/KAG技术报告.md)
65+
- [SPG白皮书](设计理念&代码架构/设计理念/SPG白皮书.md)
66+
67+
### 技术分享
68+
69+
- [KAGintroduction and applications](设计理念&代码架构/技术分享/KAGintroductionandapplications.md)
70+
- [2024.11.09ShangHai Meetup](设计理念&代码架构/技术分享/2024.11.09ShangHaiMeetup.md)
71+
72+
## 版本升级说明
73+
74+
- [开发者和产品侧差异](版本升级说明/开发者和产品侧差异.md)
75+
- [ReleaseNotes](版本升级说明/ReleaseNotes.md)
76+
77+
## 参与贡献
78+
79+
- [KAG近期迭代重点](参与贡献/KAG近期迭代重点.md)
80+
81+
## [常见问题](常见问题.md)

‎website/docs/en/DesignPhilosophy/TechReport/TechnicalReport.md

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,8 @@
1+
---
2+
sidebar_position: 2
3+
---
4+
5+
# Technical Report
6+
7+
[technical report of KAG · OpenSPG · Discussion #51](https://github.com/orgs/OpenSPG/discussions/51)
8+

‎website/docs/en/DesignPhilosophy/TechReport/WhitePaperforOpenSPG.md

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,8 @@
1+
---
2+
sidebar_position: 1
3+
---
4+
5+
# White Paper for OpenSPG
6+
7+
[white paper for openspg · OpenSPG · Discussion #50](https://github.com/orgs/OpenSPG/discussions/50)
8+

‎website/docs/en/DesignPhilosophy/TechReport/_category_.json

Lines changed: 7 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,7 @@
1+
{
2+
"label": "Tech Report",
3+
"position": 2,
4+
"link": {
5+
"type": "generated-index"
6+
}
7+
}

‎website/docs/en/DesignPhilosophy/TechnologySharing/2024.11.09ShangHaiMeetup.md

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,8 @@
1+
---
2+
sidebar_position: 1
3+
---
4+
5+
# 2024.11.09 ShangHai Meetup
6+
7+
[1109 Meetup PPT · OpenSPG · Discussion #45](https://github.com/orgs/OpenSPG/discussions/45)
8+

‎website/docs/en/DesignPhilosophy/TechnologySharing/KAGIntroductionandapplications.md

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,8 @@
1+
---
2+
sidebar_position: 2
3+
---
4+
5+
# KAG Introduction and applications
6+
7+
[KAG introduction and applications · OpenSPG · Discussion #52](https://github.com/orgs/OpenSPG/discussions/52)
8+

‎website/docs/en/DesignPhilosophy/TechnologySharing/_category_.json

Lines changed: 7 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,7 @@
1+
{
2+
"label": "Technology Sharing",
3+
"position": 1,
4+
"link": {
5+
"type": "generated-index"
6+
}
7+
}

‎website/docs/en/DesignPhilosophy/_category_.json

Lines changed: 7 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,7 @@
1+
{
2+
"label": "Design Philosophy",
3+
"position": 6,
4+
"link": {
5+
"type": "generated-index"
6+
}
7+
}

‎website/docs/en/Examples/Privatedomaindatasets/EnterpriseSupplyChainKnowledgeGraph.md

Lines changed: 10 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,10 @@
1+
---
2+
sidebar_position: 1
3+
---
4+
5+
# Enterprise Supply Chain Knowledge Graph
6+
7+
The documentation for the examples has been migrated to GitHub; please refer to
8+
9+
[KAG/kag/examples/supplychain/README.md at master · OpenSPG/KAG](https://github.com/OpenSPG/KAG/blob/master/kag/examples/supplychain/README.md)
10+

‎website/docs/en/Examples/Privatedomaindatasets/MedicalKnowledgeGraph.md

Lines changed: 11 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,11 @@
1+
---
2+
sidebar_position: 3
3+
---
4+
5+
# Medical Knowledge Graph
6+
7+
The documentation for the examples has been migrated to GitHub; please refer to
8+
9+
[KAG/kag/examples/medicine/README.md at master · OpenSPG/KAG](https://github.com/OpenSPG/KAG/blob/master/kag/examples/medicine/README.md)
10+
11+

‎website/docs/en/Examples/Privatedomaindatasets/RiskMiningKnowledgeGraph.md

Lines changed: 10 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,10 @@
1+
---
2+
sidebar_position: 2
3+
---
4+
5+
# Risk Mining Knowledge Graph
6+
7+
The documentation for the examples has been migrated to GitHub; please refer to
8+
9+
[KAG/kag/examples/riskmining/README.md at master · OpenSPG/KAG](https://github.com/OpenSPG/KAG/blob/master/kag/examples/riskmining/README.md)
10+

‎website/docs/en/Examples/Privatedomaindatasets/_category_.json

Lines changed: 7 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,7 @@
1+
{
2+
"label": "Private domain datasets",
3+
"position": 1,
4+
"link": {
5+
"type": "generated-index"
6+
}
7+
}

‎website/docs/en/Examples/Publicdomaindatasets/Hotpotqa(Multi-hopQ&A).md

Lines changed: 10 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,10 @@
1+
---
2+
sidebar_position: 2
3+
---
4+
5+
# Hotpotqa(Multi-hop Q&A)
6+
7+
The documentation for the examples has been migrated to GitHub; please refer to
8+
9+
[KAG/kag/examples/hotpotqa/README.md at master · OpenSPG/KAG](https://github.com/OpenSPG/KAG/blob/master/kag/examples/hotpotqa/README.md)
10+

‎website/docs/en/Examples/Publicdomaindatasets/Musique(Multi-hopQ&A).md

Lines changed: 10 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,10 @@
1+
---
2+
sidebar_position: 4
3+
---
4+
5+
# Musique(Multi-hop Q&A)
6+
7+
The documentation for the examples has been migrated to GitHub; please refer to
8+
9+
[KAG/kag/examples/musique/README.md at master · OpenSPG/KAG](https://github.com/OpenSPG/KAG/blob/master/kag/examples/musique/README.md)
10+

‎website/docs/en/Examples/Publicdomaindatasets/Twowiki(Multi-hopQ&A).md

Lines changed: 10 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,10 @@
1+
---
2+
sidebar_position: 3
3+
---
4+
5+
# Twowiki(Multi-hop Q&A)
6+
7+
The documentation for the examples has been migrated to GitHub; please refer to
8+
9+
[KAG/kag/examples/2wiki/README.md at master · OpenSPG/KAG](https://github.com/OpenSPG/KAG/blob/master/kag/examples/2wiki/README.md)
10+

‎website/docs/en/Examples/Publicdomaindatasets/_category_.json

Lines changed: 7 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,7 @@
1+
{
2+
"label": "Public domain datasets",
3+
"position": 2,
4+
"link": {
5+
"type": "generated-index"
6+
}
7+
}

‎website/docs/en/Examples/Publicdomaindatasets/benchmark.md

Lines changed: 57 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,57 @@
1+
---
2+
sidebar_position: 1
3+
---
4+
5+
# benchmark
6+
7+
# Performance on multi-hop factual QA tasks
8+
The EM and F1 metrics of the version V0.6 of KAG, using the same experiment configuration as the KAG technical report ([https://arxiv.org/pdf/2409.13731](https://arxiv.org/pdf/2409.13731)), are as follows:
9+
10+
| **** | **HotpotQA** | | **2Wiki** | | **MuSiQue** | **Average** |
11+
| --- | --- | --- | --- | --- | --- | --- |
12+
| | EM | F1 | EM | F1 | EM | F1 | EM | F1 |
13+
| GraphRAG | 0 | | 0 | | 0 | | 0 | |
14+
| lightRAG | 0 | | 0 | | 0 | | 0 | |
15+
| KAG TechReport<br/>（DeepSeek-V2 API） | 62.5 | 76.2 | 67.8 | 76.2 | 36.7 | 48.7 | 55.6 | 67.0 |
16+
| V0.6<br/>（DeepSeek-V2.5 API） | 60.9 | 75.4 | 69.6 | 78.6 | 36.1 | 48.2 | 55.5 | 67.4 |
17+
18+
19+
Note: The metrics of the KAG technical report are taken from Table 10 of the original text.
20+
21+
# Performance on query-focused summarization (QFS) tasks
22+
+ **Dataset**
23+
24+
To evaluate the performance of KAG on query-focused summarization tasks, we compared the outputs of KAG and LightRAG on the [UltraDomain](https://huggingface.co/datasets/TommyChien/UltraDomain/tree/main) **cs.jsonl** dataset. Please refer to [KAG Example: CSQA](https://github.com/OpenSPG/KAG/tree/master/kag/examples/csqa).
25+
26+
The **cs.jsonl** file contains 10 documents from the field of Computer Science, along with 100 questions and their corresponding answers based on these documents. Unlike the comparison method in the LightRAG paper, we used the questions provided in the **cs.jsonl** file rather than generating questions using a large language model for evaluation. Additionally, when calculating the factual correctness metric, we used the answers to the questions in the **cs.jsonl** file as the ground-truth.
27+
28+
+ **Quantitative evaluation results**
29+
30+
| **** | **KAG** | **LightRAG** |
31+
| --- | --- | --- |
32+
| Comprehensiveness（0~10） | 7.57 | 8.87 |
33+
| Diversity（0~10） | 6.87 | 8.28 |
34+
| Empowerment（0~10） | 7.54 | 8.53 |
35+
| Factual Correctness（0~1） | 0.365 | 0.352 |
36+
| Construction time consumption | 4800 seconds | 3400 seconds |
37+
| Construction token consumption | 7,006 K | 4,428 K |
38+
| Basic experiment configuration | generative model: deepseek-chat<br/>representational model: bge-m3<br/>concurrency:<br/>50 (num_threads_per_chain=50, num_chains=16)<br/>chunk size:<br/>(split_length=4950, window_length=100) | generative model: deepseek-chat<br/>representational model: bge-m3<br/>concurrency:<br/>50 (llm_model_max_async=50, embedding_func_max_async=50)<br/>chunk size:<br/>(chunk_token_size=1200, chunk_overlap_token_size=100) |
39+
40+
41+
+ **Metric Interpretation**
42+
43+
In this release, from the perspective of metrics, KAG has shown improvement in summarization tasks compared to the previous version, but there is still a gap compared to LightRAG. At the same time, in order to support both summarization generation and factual question answering, more work was done during knowledge extraction, which increased token consumption. We will continue to optimize this in future versions.
44+
45+
EM and F1 metrics are not shown in the table because when we used the HotpotQA dataset for construction and evaluation, we found that the outputs of LightRAG, GraphRAG, and KAG (using the optimized prompt) all yielded an EM of 0 and an F1 close to 0 when compared to the HotpotQA evaluation dataset. We believe that EM and F1 are not suitable for evaluating the outputs of summarization tasks.
46+
47+
48+
49+
We also acknowledge that the four evaluation metrics in the table are not perfect.
50+
51+
+ **Comprehensiveness**, **Diversity**, and **Empowerment** metrics are sensitive to the order of the answers during evaluation. Please refer to this [issue](https://github.com/HKUDS/LightRAG/issues/438) and the LightRAG paper.
52+
+ **Factual Correctness** depends on large language models, and the output is unstable. We also experimented with the Factual Correctness calculation method from RAGAS, but it was even more unstable than the method shown in the CSQA example.
53+
54+
If you have more reasonable evaluation methods, please feel free to provide feedback.
55+
56+
57+

‎website/docs/tutorial-extras/_category_.json renamed to ‎website/docs/en/Examples/_category_.json

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11
{
2-
"label": "Tutorial - Extras",
2+
"label": "Examples",
33
"position": 3,
44
"link": {
55
"type": "generated-index"

0 commit comments

Comments
 (0)