Hausa Soundex dataset with evaluation script for reproducibility

Hausa Names Soundex Dataset This dataset contains Hausa personal names, their misspellings, and associated phonetic encodings using the Sautex algorithm, a Hausa-adapted Soundex system. The dataset is intended for research in name-matching, spelling correction, and phonetic similarity in low-resource languages. Contents data/ — Contains the CSV files for the dataset. code/ — Contains the scripts for evaluation, and the Sautex phonetic encoding implementation. docs/ — Contains additional documentation and figures for evaluation results. Dataset Description The dataset was collected from spelling attempts of Hausa personal names and includes both the correct spellings and various incorrect attempts. Each record contains metadata for evaluation and reproducibility. The phonetic encodings were generated using Sautex, which adapts the Soundex rules to Hausa phonology. Data Dictionary SN Column Name Description Data Type 1 name The correct Hausa personal name String 2 attempt The spelling attempt for the name String 3 string_match Indicates if the attempt matches the intended name (True/False) Boolean 4 edit_distance Damerau-Levenshtein edit distance between attempt and name Integer 5 name_soundex_h* H* variant of the Sautex code generated for the intended name String 6 attempt_soundex_h* H* variant of the Sautex code generated for the spelling attempt String 7 match_status_h* Indicates if the H* Sautex code for attempt matches that of the name (True/False) Boolean 8 name_soundex_h** H** variant of the Sautex code generated for the intended name String 9 attempt_soundex_h** H** variant of the Sautex code generated for the spelling attempt String 10 match_status_h** Indicates if the English Soundex code for attempt matches that of the name (True/False) Boolean 11 name_soundex_eng The English Soundex code generated for the intended name String 12 attempt_soundex_eng The English Soundex code generated for the spelling attempt String 13 match_status_eng Indicates if the English Soundex code for attempt matches that of the name (True/False) Boolean 14 dataset_source Sourced from online survey using https://name-spell.thrinkle.com String Instruction To evaluate the Sautex algorithm: unzip the sautex-evaluation-python-script.zip file create a virtual environment using the command python -m venv sautexvenv. For convenience, the virtual environment should be within the sautex-evaluation-python-script folder run pip install -r requirements.txt to install project dependencies generate a report for each Sautex algorithm variant, h* and h**, and the English Soundex algorithm, eng by updating the value of the constant VARIANT on line 13 in index.py with the appropriate algorithm symbol (h*,h**,eng) in each case, run python index.py. Find the generated reports in the project folder. License This dataset is released under the CC BY 4.0 license. You are free to use, share, and adapt the dataset, provided appropriate credit is given. Citation If you use this dataset, please cite: Ephraim Bernard. Hausa Names Soundex Dataset. 2025.豪萨语姓名语音编码数据集（Hausa Names Soundex Dataset） 本数据集收录豪萨语人名及其拼写错误变体，以及采用适配豪萨语语音系统的Sautex算法（一种针对豪萨语改造的Soundex语音编码系统）生成的关联语音编码。本数据集旨在面向低资源语言领域的姓名匹配、拼写校正及语音相似度相关研究。 ## 内容说明 - `data/`：存放本数据集的CSV格式文件 - `code/`：存放评估脚本及Sautex语音编码实现代码 - `docs/`：存放额外的说明文档及评估结果相关图表 ## 数据集说明 本数据集采集自豪萨语人名的拼写尝试数据，涵盖正确拼写形式与各类错误拼写变体。每条数据均包含用于评估与可复现研究的元数据。 本数据集的语音编码通过Sautex算法生成，该算法将标准Soundex规则适配至豪萨语语音学体系。 ## 数据字典 | 序号 | 列名 | 字段说明 | 数据类型 | |------|--------------------------|--------------------------------------------------------------------------|----------| | 1 | `name` | 正确的豪萨语人名 | 字符串 | | 2 | `attempt` | 该姓名的拼写尝试版本 | 字符串 | | 3 | `string_match` | 标识拼写尝试是否与目标姓名完全匹配（取值为`True`/`False`） | 布尔型 | | 4 | `edit_distance` | 拼写尝试与正确姓名之间的达梅罗-莱文施泰因（Damerau-Levenshtein）编辑距离 | 整数型 | | 5 | `name_soundex_h*` | 为目标姓名生成的Sautex编码H*变体 | 字符串 | | 6 | `attempt_soundex_h*` | 为拼写尝试生成的Sautex编码H*变体 | 字符串 | | 7 | `match_status_h*` | 标识拼写尝试的H*型Sautex编码是否与目标姓名的H*型Sautex编码一致（取值为`True`/`False`） | 布尔型 | | 8 | `name_soundex_h**` | 为目标姓名生成的Sautex编码H**变体 | 字符串 | | 9 | `attempt_soundex_h**` | 为拼写尝试生成的Sautex编码H**变体 | 字符串 | | 10 | `match_status_h**` | 标识拼写尝试的英语Soundex编码是否与目标姓名的英语Soundex编码一致（取值为`True`/`False`） | 布尔型 | | 11 | `name_soundex_eng` | 为目标姓名生成的标准英语Soundex编码 | 字符串 | | 12 | `attempt_soundex_eng` | 为拼写尝试生成的标准英语Soundex编码 | 字符串 | | 13 | `match_status_eng` | 标识拼写尝试的英语Soundex编码是否与目标姓名的英语Soundex编码一致（取值为`True`/`False`） | 布尔型 | | 14 | `dataset_source` | 数据采集自https://name-spell.thrinkle.com在线调研平台 | 字符串 | ## 评估流程 如需评估Sautex算法，请遵循以下步骤： 1. 解压`sautex-evaluation-python-script.zip`压缩包 2. 创建虚拟环境，执行命令：`python -m venv sautexvenv`。为便于管理，建议将虚拟环境创建于`sautex-evaluation-python-script`文件夹内 3. 执行`pip install -r requirements.txt`安装项目依赖项 4. 如需生成对应算法变体（`h*`、`h**`或`eng`）的评估报告，请修改`index.py`文件第13行的常量`VARIANT`的取值为对应算法标识，随后执行`python index.py`。生成的评估报告将存储于项目文件夹中。 ## 授权协议 本数据集采用CC BY 4.0协议发布。您可自由使用、分享及改编本数据集，但需注明原作者与来源。 ## 引用方式 如您使用本数据集，请引用如下文献： > Ephraim Bernard. 豪萨语姓名语音编码数据集. 2025.