boun-tabilab/turkish_parliamentary_data

--- license: cc-by-nc-sa-4.0 language: - tr - ota pretty_name: Grand National Assembly Corpus of Türkiye (GNACT) size_categories: - 1M<n<10M task_categories: - text-generation - fill-mask - text-classification - question-answering tags: - parliamentary - turkish - ottoman-turkish - historical - political-discourse - ocr configs: - config_name: full_sessions data_files: - split: train path: full_sessions/*.parquet - config_name: pages data_files: - split: train path: pages/*.parquet - config_name: tbmm_only data_files: - split: train path: tbmm_only/*.parquet --- # Grand National Assembly Corpus of Türkiye (GNACT) A comprehensive collection of Turkish parliamentary transcripts spanning over 100 years (1920–present), from 10 legislative bodies. Includes both Ottoman Turkish (1920–1928) and Modern Turkish (1928–present) texts. ## Loading the dataset ```python from datasets import load_dataset # Strategy 1: full session documents, all bodies (default) ds = load_dataset("boun-tabilab/turkish_parliamentary_data", "full_sessions", split="train") # Strategy 2: page-level, all bodies ds = load_dataset("boun-tabilab/turkish_parliamentary_data", "pages", split="train") # Strategy 3: TBMM only, all terms ds = load_dataset("boun-tabilab/turkish_parliamentary_data", "tbmm_only", split="train") # Filter after loading (e.g. TBMM term 22, year 3) ds_t22 = ds.filter(lambda x: x["term"] == 22) ds_t22_y3 = ds.filter(lambda x: x["term"] == 22 and x["year"] == 3) ``` ## Configs | Config | Granularity | Bodies | Shards | |---|---|---|---| | `full_sessions` | Document | All 10 | 27 | | `pages` | Page | All 10 | 20 | | `tbmm_only` | Document | TBMM only | 16 | ## Columns | Column | Type | Description | |---|---|---| | `document_id` | string | Unique document identifier | | `legislative_body` | string | Legislative body name | | `term` | int32 | Legislative term number | | `year` | int32 | Year within the term | | `session` | int32 | Session number | | `volume` | int32 | Volume number | | `page_num` | int32 | Page number (0 for full-session rows) | | `text` | string | OCR-extracted text | | `is_full_session` | bool | True for document rows, False for page rows | | `language` | string | `tr` (Modern Turkish) or `ota` (Ottoman Turkish) | ## Legislative bodies | Code | Name | |---|---| | TBMM | Türkiye Büyük Millet Meclisi (Grand National Assembly) | | MM | Millet Meclisi (National Assembly) | | CS | Cumhuriyet Senatosu (Senate of the Republic) | | BT | Birleşik Toplantı (Joint Sessions) | | MGK | Milli Güvenlik Konseyi (National Security Council) | | MBK | Milli Birlik Komitesi (National Unity Committee) | | KM | Kurucu Meclis (Constituent Assembly) | | DM | Danışma Meclisi (Advisory Council) | | TM | Temsilciler Meclisi (House of Representatives) | | GC | Gizli Celse / Kapalı Oturum (Secret/Closed Sessions) | ## License [CC BY-NC-SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/) ## Citation ```bibtex @dataset{turkish_parliamentary_data_2026, title = {Turkish Parliamentary Data 2026: Grand National Assembly Corpus of Türkiye (GNACT)}, year = {2026}, url = {https://huggingface.co/datasets/boun-tabilab/turkish_parliamentary_data}, license = {CC BY-NC-SA 4.0} } ```--- license: 知识共享署名-非商业性使用-相同方式共享4.0国际许可协议（CC BY-NC-SA 4.0） language: - tr（土耳其语） - ota（奥斯曼土耳其语，Ottoman Turkish） pretty_name: 土耳其大国民议会语料库（Grand National Assembly Corpus of Türkiye，缩写GNACT） size_categories: - 100万<样本量<1000万 task_categories: - 文本生成 - 掩码填空 - 文本分类 - 问答 tags: - 议会相关 - 土耳其语 - 奥斯曼土耳其语 - 历史文本 - 政治话语 - 光学字符识别（Optical Character Recognition，OCR） configs: - config_name: full_sessions data_files: - split: train path: full_sessions/*.parquet - config_name: pages data_files: - split: train path: pages/*.parquet - config_name: tbmm_only data_files: - split: train path: tbmm_only/*.parquet --- # 土耳其大国民议会语料库（GNACT） 本数据集为涵盖10个立法机构、跨度超百年（1920年至今）的土耳其议会议事记录综合语料库，包含1920至1928年的奥斯曼土耳其语（Ottoman Turkish）文本与1928年至今的现代土耳其语文本。 ## 数据集加载方式 python from datasets import load_dataset # 方案1：完整议事文档，包含全部立法机构（默认方案） ds = load_dataset("boun-tabilab/turkish_parliamentary_data", "full_sessions", split="train") # 方案2：页面级粒度，包含全部立法机构 ds = load_dataset("boun-tabilab/turkish_parliamentary_data", "pages", split="train") # 方案3：仅包含土耳其大国民议会（TBMM），涵盖全部届次 ds = load_dataset("boun-tabilab/turkish_parliamentary_data", "tbmm_only", split="train") # 加载后自定义筛选（例如筛选第22届届次、第3年的数据） ds_t22 = ds.filter(lambda x: x["term"] == 22) ds_t22_y3 = ds.filter(lambda x: x["term"] == 22 and x["year"] == 3) ## 配置项 | 配置名称 | 数据粒度 | 涵盖立法机构 | 分片数 | |---|---|---|---| | `full_sessions` | 文档级 | 全部10个 | 27 | | `pages` | 页面级 | 全部10个 | 20 | | `tbmm_only` | 文档级 | 仅土耳其大国民议会（TBMM） | 16 | ## 数据列说明 | 列名 | 数据类型 | 说明 | |---|---|---| | `document_id` | string | 唯一文档标识符 | | `legislative_body` | string | 立法机构名称 | | `term` | int32 | 立法届次编号 | | `year` | int32 | 届次内年份 | | `session` | int32 | 议事次数 | | `volume` | int32 | 卷号 | | `page_num` | int32 | 页码（完整议事文档行该字段取值为0） | | `text` | string | 光学字符识别（Optical Character Recognition，OCR）提取的文本 | | `is_full_session` | bool | 文档行取值为True，页面级行取值为False | | `language` | string | `tr`（现代土耳其语）或`ota`（奥斯曼土耳其语，Ottoman Turkish） | ## 立法机构列表 | 机构代码 | 机构名称 | |---|---| | TBMM | 土耳其大国民议会（Türkiye Büyük Millet Meclisi，Grand National Assembly） | | MM | 国民议会（Millet Meclisi，National Assembly） | | CS | 共和国参议院（Cumhuriyet Senatosu，Senate of the Republic） | | BT | 联合议事会（Birleşik Toplantı，Joint Sessions） | | MGK | 国家安全委员会（Milli Güvenlik Konseyi，National Security Council） | | MBK | 民族团结委员会（Milli Birlik Komitesi，National Unity Committee） | | KM | 制宪会议（Kurucu Meclis，Constituent Assembly） | | DM | 咨询委员会（Danışma Meclisi，Advisory Council） | | TM | 众议院（Temsilciler Meclisi，House of Representatives） | | GC | 秘密/闭门议事会（Gizli Celse / Kapalı Oturum，Secret/Closed Sessions） | ## 许可证 [知识共享署名-非商业性使用-相同方式共享4.0国际许可协议（CC BY-NC-SA 4.0）](https://creativecommons.org/licenses/by-nc-sa/4.0/) ## 引用格式 bibtex @dataset{turkish_parliamentary_data_2026, title = {Turkish Parliamentary Data 2026: Grand National Assembly Corpus of Türkiye (GNACT)}, year = {2026}, url = {https://huggingface.co/datasets/boun-tabilab/turkish_parliamentary_data}, license = {CC BY-NC-SA 4.0} }