File size: 2,991 Bytes
7ec374c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
---
license: apache-2.0
language:
- en
tags:
- software-engineering
- conventional-commits
- nlp
pretty_name: CCSBENCH
size_categories:
- 10K<n<100K
---

# CCSBENCH: A Fine-Grained Benchmark for Conventional Commits

## 1. Dataset Summary
CCSBENCH is a large-scale benchmark dataset specifically constructed for fine-grained **Conventional Commits (CCS)** classification research. Based on the CCS specification, it provides high-quality commit data with verified types and scopes across multiple programming languages.

The dataset contains **93,807** CCS-compliant commits extracted from 119 high-quality repositories. It spans a wide temporal range (April 2017 to February 2023) and includes **142 distinct commit types** and **6,633 unique scopes**.

## 2. Dataset Structure

### Data Splits
We employ a strict **temporal split** to simulate real-world scenarios, ensuring that repositories are tracked continuously over time.

| File | Count | Usage |
| :--- | :--- | :--- |
| `train.json` | 74,581 | Model training |
| `valid.json` | 9,939 | Hyperparameter tuning |
| `test.json` | 9,287 | Final performance evaluation |

### Language Distribution (by Repository)
The benchmark covers 9 key programming languages:

| Language | Repositories | Language | Repositories |
| :--- | :--- | :--- | :--- |
| **TypeScript** | 61 | **Rust** | 2 |
| **JavaScript** | 34 | **Java** | 2 |
| **Go** | 11 | **C++** | 1 |
| **Python** | 4 | **Kotlin** | 1 |
| **C#** | 3 | | |

## 3. Data Schema
Each record in the dataset is a JSON object representing a single commit. Below is a description of the key fields:

| Field | Type | Description |
| :--- | :--- | :--- |
| `hash` | string | Unique identifier for the commit. |
| `message` | string | The full original commit message. |
| `commit_type` | string | The extracted CCS type (e.g., `feat`, `fix`, `chore`). |
| `commit_scope` | string | The extracted CCS scope (can be `null` if not provided). |
| `is_CCS` | integer | Binary indicator (1) for CCS compliance. |
| `language` | string | Programming language of the repository. |
| `repo` | string | Full name of the source repository (Owner/Repo). |
| `mods` | array | List of modified files, including `diff`, `new_path`, and `old_path`. |
| `date` | string | Timestamp of the commit. |
| `license` | string | Original license of the source repository. |

## 4. Construction Principles
The construction of CCSBENCH prioritizes **Scale, Quality, and Repository Continuity**:
1. **Keyword Detection**: Identifying repositories explicitly adopting CCS ("conventionalcommits.org").
2. **Format Validation**: Using a robust parser to validate `<type>[optional scope]: <description>`.
3. **Temporal Alignment**: Discarding all commits preceding the repository’s formal adoption of CCS to ensure data purity.

## 5. Licensing
- **Dataset License**: **Apache-2.0** (compatible with the licenses of the original repositories in CommitChronicle).
- **Code License**: **MIT** (for processing scripts).