|
--- |
|
license: mit |
|
task_categories: |
|
- text-generation |
|
- table-question-answering |
|
language: |
|
- ru |
|
- en |
|
tags: |
|
- 1с |
|
- ones |
|
- 1C |
|
- 1С |
|
- один-с |
|
pretty_name: 1C_forums_dataset_by_arefaste |
|
size_categories: |
|
- 1K<n<10K |
|
configs: |
|
- config_name: default |
|
data_files: |
|
- split: train |
|
path: result_final_make_dataset.parquet |
|
--- |
|
# 1C_forums: Dataset based parsed data from two of the most popular forums for 1c |
|
|
|
This dataset made of the parsed data from two forums for coders at the 1C languages: |
|
- [Infostart](https://forum.infostart.ru/group2/?a=26694) - all threads presents as like think section for model, and marked as the best result for queestion as final answer |
|
- [Fastcode](https://fastcode.im/Templates?TemplatesOnly=True) - Only templates |
|
|
|
## Dataset Overview |
|
All rows prepared as useful columns. All text prepared as markdown text, and code 1c looks as like: |
|
```md |
|
\`\`\`1c |
|
"ВЫБРАТЬ |
|
| Ссылка.Контрагент, |
|
| Ссылка.Магазин, |
|
| СУММА(КоличествоСпрос) КАК КолвоСпрос |
|
|ИЗ |
|
| Документ.СпросНаТовар.Товары |
|
|ГДЕ Ссылка.Дата МЕЖДУ ДОБАВИТЬКДАТЕ(&ДатаНачало, ГОД, -10) И &ДатаНачало |
|
| СГРУППИРОВАТЬ ПО Ссылка.Контрагент, СсылкаМагазин";. |
|
\`\`\` |
|
``` |
|
|
|
### ❗IMPORTANT❗: Escaping special characters |
|
|
|
Values in columns `prompt`, `think_process`, and `solution` have escaped special characters like `\n`, `\t`, and `\\`! |
|
|
|
The reason for this behavior: the Python's `csv` module that I'm using in the parser to save rows saves `\n` as an actual newline character, |
|
which makes reading the prepared file with Pandas impossible. |
|
|
|
To fix it use simple lambda code: |
|
```python |
|
return_back_special_symbols = lambda x: x.replace("\\n", "\n").replace("\\t", "\t").replace("\\\\", "\\") |
|
``` |
|
|
|
|
|
|
|
### Infostart |
|
Choosen data at [Infostart](https://forum.infostart.ru/group2/?a=26694) are all forums threads with found answers in section Dev. |
|
- `source` - for all rows for Infostart forum contains this value `forum_infostart` |
|
- `in_source_id` - id of the thread from that web forum. To get link to this page use this format: `https://forum.infostart.ru/forum9/{in_source_id}/` |
|
- `prompt` - it's a first question(msg) in thread. |
|
- `think_process` - its all thread discution presented as like think process that in there each msg between others separeted `\n----\n` |
|
- `solution` - msgs marked in thread as like solution for question |
|
- `is_answer_a_link` - if link contains a more then 60-70% of all solution, it's contains value `True`, else `False` |
|
- `has_link` - Counter links in solution. If solution doesn't contains any link, it value will be a `Nan` |
|
- `tags` - predicted tags |
|
|
|
|
|
### FastCode |
|
All data at [Fastcode](https://fastcode.im/Templates?TemplatesOnly=True) are only templates, comments for this code made as new section `# Примечание` that what shows each comment. |
|
- `source` - for all rows for Fastcode forum contains this value `fastcode_Templates` |
|
- `in_source_id` - id of the thread from that web forum. To get link to this page use this format: `https://fastcode.im/Templates/{in_source_id}/` |
|
- `prompt` - it's title of the template. |
|
- `think_process` - **`None` - it's dont use it at all!** |
|
- `solution` - it's description(if it's contains there) and code with notes(it's comments, if its are there) |
|
- `is_answer_a_link` - if link contains a more then 60-70% of all solution, it's contains value `True`, else `False` |
|
- `has_link` - Counter links in solution. If solution doesn't contains any link, it value will be a `Nan` |
|
- `tags` - Tags at template separated by comma |
|
|
|
## Dataset Statistics |
|
|
|
* **Total examples**: 19041 |
|
- forum ***infostart***: 18299 |
|
- ***fastcode*** Template: 742 |
|
|
|
## Parsing scripts |
|
|
|
You can find in folder `scripts_parsing`! |
|
|
|
## Citation |
|
|
|
Please cite this dataset if you use it in your work: |
|
|
|
``` |
|
@misc{arefaste2025, |
|
title={Arefaste: Parsed dataset for 1c from dataset}, |
|
author={arefaste}, |
|
year={2025}, |
|
publisher={Hugging Face}, |
|
howpublished={\url{https://huggingface.co/datasets/arefaste/1C_Forums}} |
|
} |
|
``` |