IRM_chat_all

## 简介

**IRM_chat_little** 数据集由 **YKDuan** 在 ModelScope 平台上贡献。该数据集专为图书情报学领域的对话式交互而设计。

**数据来源：**
- 大百科全书中关于图书情报学科的概念
- CSSCI（中文社会科学引文索引）中图书情报学科的题录信息

这些基础文本通过大型语言模型（LLMs）构建了总计 **276,083 条对话轮次**。该数据集非常适合用于学术对话领域的 LLM 训练和微调，特别是在图书情报学领域。

## 数据集内容

**IRM_chat_little** 数据集包含 **276,083 条对话轮次**。每个条目都以 JSON 对象的形式呈现，代表用户和助手之间的对话。

### 格式

```json
{"messages": [{"role": "user", "content": "用户查询"}, {"role": "assistant", "content": "助手回复"}]}
```

### 示例

```json
{"messages": [{"role": "user", "content": "什么是信息检索？"}, {"role": "assistant", "content": "信息检索（IR）是根据信息需求从信息资源集合中获取相关信息系统资源的行为。搜索可以基于元数据或全文索引。"}]}
```

## 数据集统计

- **对话轮次总数：** 276,083 个独特条目

## 使用方法

该数据集可以通过 ModelScope 库进行访问和使用。

### 使用 ModelScope SDK (Python)

```python
from modelscope.msdatasets import MsDataset

# 加载数据集
dataset = MsDataset.load('YKDuan/IRM_chat_little')

# 迭代数据集
for item in dataset:
print(item)
break # 打印第一个条目以供演示
```

## 许可协议

该数据集根据 **Apache 2.0** 许可协议分发。请参考 ModelScope 平台以获取最新的许可信息。

## 引用

如果您在研究或应用中使用此数据集，请引用以下论文：

```bibtex
@inproceedings{Zhu2025LISGPT,
title={{LISGPT: Boundary Knowledge Enhanced Academic Large Language Modeling and its Scenario Applications}},
author={Zhu, Y and Duan, Y* and Hu, J and Jin, J and Ye, J},
booktitle={Proceedings of ASIS\&T 2025},
year={2025}
}
```

## 联系方式

有关此数据集的任何问题或疑虑，请联系 **[e-mail]** 或参考 ModelScope 社区。

## Introduction

The **IRM_chat_little** dataset was contributed by **YKDuan** on the ModelScope platform. This dataset is specifically designed for conversational interactions in the field of library and information science (LIS).

## Data Sources
- Concepts related to library and information science from general encyclopedias
- Bibliographic records of library and information science publications indexed in CSSCI (Chinese Social Sciences Citation Index)

These base texts were used to construct a total of **276,083 conversational turns** via large language models (LLMs). This dataset is highly suitable for LLM training and fine-tuning in the domain of academic conversations, particularly within library and information science.

## Dataset Content

The **IRM_chat_little** dataset contains **276,083 conversational turns**. Each entry is presented as a JSON object representing a conversation between a user and an assistant.

### Format
json
{"messages": [{"role": "user", "content": "user query"}, {"role": "assistant", "content": "assistant reply"}]}

### Example
json
{"messages": [{"role": "user", "content": "什么是信息检索？"}, {"role": "assistant", "content": "信息检索（IR）是根据信息需求从信息资源集合中获取相关信息系统资源的行为。搜索可以基于元数据或全文索引。"}]}

## Dataset Statistics
- **Total Conversational Turns**: 276,083 unique entries

## Usage Instructions

This dataset can be accessed and utilized via the ModelScope library.

### Using ModelScope SDK (Python)
python
from modelscope.msdatasets import MsDataset

# Load the dataset
dataset = MsDataset.load('YKDuan/IRM_chat_little')

# Iterate through the dataset
for item in dataset:
print(item)
break # Print the first entry for demonstration

## License Agreement

This dataset is distributed under the **Apache 2.0** license. Please refer to the ModelScope platform for the latest licensing information.

## Citation

If you use this dataset in your research or applications, please cite the following paper:
bibtex
@inproceedings{Zhu2025LISGPT,
title={{LISGPT: Boundary Knowledge Enhanced Academic Large Language Modeling and its Scenario Applications}},
author={Zhu, Y and Duan, Y* and Hu, J and Jin, J and Ye, J},
booktitle={Proceedings of ASIS&T 2025},
year={2025}
}

## Contact Information

For any questions or concerns regarding this dataset, please contact **[e-mail]** or refer to the ModelScope community.