README.md

---
configs:
- config_name: DDC
  default: true
  data_files:
  - split: train
    path:
    - DDC.jsonl.gz
- config_name: LCC
  data_files:
  - split: train
    path:
    - LCC.jsonl.gz
- config_name: UDC
  data_files:
  - split: train
    path:
    - UDC.jsonl.gz
- config_name: CLC
  data_files:
  - split: train
    path:
    - CLC.jsonl.gz
- config_name: RVK
  data_files:
  - split: train
    path:
    - RVK.jsonl.gz
- config_name: BBK
  data_files:
  - split: train
    path:
    - BBK.jsonl.gz
task_categories:
- text-classification
task_ids:
- multi-class-classification
pretty_name: Library Classification Systems
tags:
- library-science
- information-organization
- ontology
language:
- en
- zh
- de
- ru
---

# Library Classification Systems

This comprehensive dataset contains hierarchical outlines of major library classification systems, offering a valuable resource for researchers, librarians, and information scientists.

| Classification System | Abbreviation | Primary Usage | Language | Entries |
|-----------------------|--------------|----------------|----------|---------|
| Dewey Decimal Classification | DDC | International | English | 1110 |
| Library of Congress Classification | LCC | International | English | 6517 |
| Universal Decimal Classification | UDC | International | English | 2431 |
| Chinese Library Classification | CLC | China | Simplified Chinese | 8826 |
| Regensburger Verbundklassifikation | RVK | German universities | German | 5032 |
| Библиотечно-библиографическая классификация | BBK | Russia | Russian | 2588 |

**Total Entries:** 26504

Note that the full classification systems can be much more detailed than presented here. For DDC, LCC, and UDC, they have to be purchased from their organizations.

## Data Structure

Each instance in the dataset represents a library classification entry.

The data is structured in **JSONL (JSON Lines)** format, with each entry containing the following fields:

| Field        | Type              | Description                                                  | Example                                      |
|--------------|-------------------|--------------------------------------------------------------|----------------------------------------------|
| `call_number`| string            | The alphanumeric call number that uniquely identifies the classification | `"AC1-999"`                                 |
| `description`| string            | A textual description of the classification topic or subject area | `"Collections. Series. Collected works"`    |
| `broader`    | string or null    | The call number of the parent/broader classification; null for root-level classifications | `null` (for a top-level classification)     |
| `narrower`   | list of strings   | An array of call numbers representing child/narrower classifications | `["AC1-195", "AC200", "AC801-895", "AC901-995", "AC999"]` |

### Example Instance

```json
{
  "call_number": "AC1-999",
  "description": "Collections. Series. Collected works",
  "broader": null,
  "narrower": ["AC1-195", "AC200", "AC801-895", "AC901-995", "AC999"]
}
```

## Dataset Creation
<details>
<summary>Click for details</summary>

On 2024-09-18, publicly available classification outlines were collected from:

- **LCC**: Library of Congress Classification Outline
- **UDC**: Universal Decimal Classification Summary (ROSSIO Vocabulary Server)
- **DDC**: Dewey Services - Resources (OCLC)
- **CLC**: CLC Index
- **RVK**: RVK Online
- **BBK**: КлассИнформ (crawled 2024-09-19)

Versions used:
- UDC Master Reference File (2011)
- DDC 23 Summaries
- RVK aktuelle csv-Datei (2023_2)

Data processing steps:
1. **DDC**: Downloaded summary PDF and formatted for consistency.
2. **LCC**: Converted Word files to plain text, extracting headings and call numbers programmatically.
3. **UDC**: Parsed `udc-summary.rdf` using RDFlib, retaining only English labels.
4. **CLC**: Crawled website, including only the first 3 hierarchy levels.
5. **RVK**: Processed CSV file, keeping only the first 3 hierarchy levels.
6. **BBK**: Crawled website, no limit on hierarchy levels.
</details>

## Supported Tasks and Leaderboards

This dataset is valuable for various tasks in information science and machine learning, including:

1. **Hierarchical Text Classification**
2. **Library Science and Information Organization**
3. **Ontology and Knowledge Graph Construction**
4. **Comparative Studies of Classification Systems**
5. **Machine Learning for Library Cataloging**

While no leaderboards are currently associated with this dataset, it provides opportunities for developing benchmarks in automated classification accuracy and ontology mapping.

## Considerations for Using the Data

### Social Impact

This dataset broadens access to library classification knowledge, potentially:

- Facilitating research in information science and digital humanities.
- Supporting improved information retrieval systems.
- Enhancing educational resources for library and information science students.

### Discussion of Biases

Be aware that library classification systems may reflect historical, cultural, and geographical biases, such as:

- **Overrepresentation of Western Perspectives**
- **Outdated Terminology or Categorizations**
- **Uneven Depth of Coverage** across subject areas.

### Known Limitations

1. **Scope**: Includes only freely available outlines, not complete texts.
2. **Granularity**: Detail levels vary across systems and subjects.
3. **Currency**: Updates to classification systems may not be immediately reflected.
4. **Structural Differences**: UDC's structure differs from DDC and LCC, affecting comparisons.

## Additional Information

### Dataset Curator

Alan Tseng

### Licensing Information

Creative Commons Attribution 4.0 International (CC BY 4.0)