Add database_clear as a denoised control environment
Browse filesProvide aligned, noise-removed counterparts of assets/database so evaluations can isolate how data friction affects measured model capability.
- README.md +5 -1
- README.zh-CN.md +4 -0
- assets/database_clear/enterprise/company_core.csv +3 -0
- assets/database_clear/enterprise/company_operation_status.csv +3 -0
- assets/database_clear/enterprise/company_profile.csv +3 -0
- assets/database_clear/enterprise/company_translation_english_chinese.json +0 -0
- assets/database_clear/industry/national_industry_status.csv +3 -0
- assets/database_clear/industry/regional_industry_status.csv +3 -0
- assets/database_clear/internal_metrics.csv +3 -0
- assets/database_clear/policy/policy_release_status.csv +3 -0
- assets/database_clear/policy/policy_resource.csv +3 -0
README.md
CHANGED
|
@@ -71,7 +71,9 @@ DataClawBench simulates at scale the noisy, weakly-semantic, cross-domain data e
|
|
| 71 |
|
| 72 |
Key directories and scripts:
|
| 73 |
|
| 74 |
-
- `assets/database/`: Benchmark data files, injected wholesale into the container workspace at run time. The root contains
|
|
|
|
|
|
|
| 75 |
- `assets/qa_raw/`: Raw task source files.
|
| 76 |
- `assets/qa_gold/`: Minimized gold files derived from `qa_raw`.
|
| 77 |
- `tasks/`: Generated OpenClaw task spec files.
|
|
@@ -362,6 +364,8 @@ Under a theme-domain view and business-oriented taxonomy, the current data envir
|
|
| 362 |
|
| 363 |
> At execution time, agents typically need to align entities across files, join across tables, normalize definitions, and perform aggregation calculations, rather than simply looking up values in a single file; when needed, they must also consult business conventions in `internal_metrics.csv`. This is the core value of DataClawBench for evaluating real-scenario data understanding and reasoning capabilities.
|
| 364 |
|
|
|
|
|
|
|
| 365 |
### 📋 Task Statistics
|
| 366 |
|
| 367 |
The current version contains **492** tasks across **7** categories, with an overall difficulty distribution of **131 easy / 286 medium / 75 hard**.
|
|
|
|
| 71 |
|
| 72 |
Key directories and scripts:
|
| 73 |
|
| 74 |
+
- `assets/database/`: Benchmark data files, injected wholesale into the container workspace at run time. The root contains
|
| 75 |
+
`internal_metrics.csv` (internal business-logic knowledge base); `enterprise/`, `industry/`, and `policy/` hold the three theme-domain datasets.
|
| 76 |
+
- `assets/database_clear/`: Cleaned counterpart of `assets/database/`. Built from the original data after completing data alignment and removing noise; provided as a control environment for measuring how the data environment affects model capability.
|
| 77 |
- `assets/qa_raw/`: Raw task source files.
|
| 78 |
- `assets/qa_gold/`: Minimized gold files derived from `qa_raw`.
|
| 79 |
- `tasks/`: Generated OpenClaw task spec files.
|
|
|
|
| 364 |
|
| 365 |
> At execution time, agents typically need to align entities across files, join across tables, normalize definitions, and perform aggregation calculations, rather than simply looking up values in a single file; when needed, they must also consult business conventions in `internal_metrics.csv`. This is the core value of DataClawBench for evaluating real-scenario data understanding and reasoning capabilities.
|
| 366 |
|
| 367 |
+
**Cleaned control environment (`assets/database_clear/`).** In addition to the default noisy environment above, this repository provides `assets/database_clear/`: a cleaned counterpart of the same sources after data alignment and noise removal. Comparing evaluation results on `database` vs. `database_clear` helps isolate how much of the observed performance is attributable to environmental friction rather than to the model itself.
|
| 368 |
+
|
| 369 |
### 📋 Task Statistics
|
| 370 |
|
| 371 |
The current version contains **492** tasks across **7** categories, with an overall difficulty distribution of **131 easy / 286 medium / 75 hard**.
|
README.zh-CN.md
CHANGED
|
@@ -48,6 +48,7 @@ DataClawBench 大规模地模拟了高噪声、弱语义、跨领域的真实世
|
|
| 48 |
关键目录与脚本职责如下:
|
| 49 |
|
| 50 |
- `assets/database/`:评测数据文件,运行时会整体注入到容器工作区。根目录含 `internal_metrics.csv`(业务逻辑内部知识库);`enterprise/`、`industry/`、`policy/` 为三大主题域数据。
|
|
|
|
| 51 |
- `assets/qa_raw/`:原始任务源文件。
|
| 52 |
- `assets/qa_gold/`:由 `qa_raw` 归约得到的精简 gold 文件。
|
| 53 |
- `tasks/`:生成后的 OpenClaw task 规范文件。
|
|
@@ -338,6 +339,9 @@ DataClawBench 数据并非来源于合成样本或教学示例,而是基于发
|
|
| 338 |
|
| 339 |
> 在任务执行层面,智能体通常需要在多文件间完成实体对齐、跨表关联、口径归一与聚合计算,而非单文件查值;必要时还需结合 `internal_metrics.csv` 中的业务约定。这也是 DataClawBench 用于评估真实场景数据理解与推理能力的核心价值。
|
| 340 |
|
|
|
|
|
|
|
|
|
|
| 341 |
### 📋 任务统计信息
|
| 342 |
|
| 343 |
当前版本共 492 个任务,覆盖 7 个类别;整体难度分布为 131 easy / 286 medium / 75 hard。
|
|
|
|
| 48 |
关键目录与脚本职责如下:
|
| 49 |
|
| 50 |
- `assets/database/`:评测数据文件,运行时会整体注入到容器工作区。根目录含 `internal_metrics.csv`(业务逻辑内部知识库);`enterprise/`、`industry/`、`policy/` 为三大主题域数据。
|
| 51 |
+
- `assets/database_clear/`:在 `assets/database/` 基础上完成数据对齐、去除数据噪声后的干净数据,用作对照环境,便于测定数据环境对模型能力的影响。
|
| 52 |
- `assets/qa_raw/`:原始任务源文件。
|
| 53 |
- `assets/qa_gold/`:由 `qa_raw` 归约得到的精简 gold 文件。
|
| 54 |
- `tasks/`:生成后的 OpenClaw task 规范文件。
|
|
|
|
| 339 |
|
| 340 |
> 在任务执行层面,智能体通常需要在多文件间完成实体对齐、跨表关联、口径归一与聚合计算,而非单文件查值;必要时还需结合 `internal_metrics.csv` 中的业务约定。这也是 DataClawBench 用于评估真实场景数据理解与推理能力的核心价值。
|
| 341 |
|
| 342 |
+
**干净对照环境(`assets/database_clear/`)。** 除上述默认的高噪声数据环境外,本仓库还提供 `assets/database_clear/`:在原数据基础上完成数据对齐并去除噪声后的干净数据。对比 `database` 与 `database_clear` 两套环境上的评测结果,便于测定数据环境对模型能力的影响,将环境摩擦与模型能力本身区分开来。
|
| 343 |
+
|
| 344 |
+
|
| 345 |
### 📋 任务统计信息
|
| 346 |
|
| 347 |
当前版本共 492 个任务,覆盖 7 个类别;整体难度分布为 131 easy / 286 medium / 75 hard。
|
assets/database_clear/enterprise/company_core.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d5dca195a01a7b368fde4af6f8284c34f723147bcaf8e3527bfe02aecf718e82
|
| 3 |
+
size 18714586
|
assets/database_clear/enterprise/company_operation_status.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4af094ff21fba8c4d856de7a0b2c3159d0af0a4bd70cc36b4faff0930e75486b
|
| 3 |
+
size 1781476
|
assets/database_clear/enterprise/company_profile.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ecd68737f4f453746c3271b146ef5f6b6e741536aab4fc8ae2a32be89085e3c2
|
| 3 |
+
size 1783742
|
assets/database_clear/enterprise/company_translation_english_chinese.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
assets/database_clear/industry/national_industry_status.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fc0853e0ecde4d4fe73d128d79fbdc7930bed9b57cfc718c68f554c2cea7517a
|
| 3 |
+
size 79920
|
assets/database_clear/industry/regional_industry_status.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ad07d2990eea08b4aaa914588f7856461ea3c3de289b72176501756a937e079b
|
| 3 |
+
size 1013131
|
assets/database_clear/internal_metrics.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f584fb7cacf19fcdc50e59b398fc65ad041d9651860d9cfe1d7afb277dbeb9dc
|
| 3 |
+
size 2238
|
assets/database_clear/policy/policy_release_status.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a784a9301e6a8feb63d7b5e2fae7da0f4a3fee3d1871703cf1560d36163184ef
|
| 3 |
+
size 1109101
|
assets/database_clear/policy/policy_resource.csv
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8322256232d053b49b7b0e7f842e811224a203f07bf51d1c890c62bdb6ea1a6f
|
| 3 |
+
size 20833538
|