ysginc/wordnet

---
pretty_name: "WordNet 3.0"
license: "other"
tags:
- "lexicon"
- "wordnet"
- "nlp"
- "datasets"
- "tabular"
configs:
- config_name: "aggregate"
default: true
data_files:
- split: "noun"
path: "noun.parquet"
- split: "verb"
path: "verb.parquet"
- split: "adj"
path: "adj.parquet"
- split: "adv"
path: "adv.parquet"
- config_name: "data"
data_files:
- split: "noun"
path: "variants/data/noun.parquet"
- split: "verb"
path: "variants/data/verb.parquet"
- split: "adj"
path: "variants/data/adj.parquet"
- split: "adv"
path: "variants/data/adv.parquet"
- config_name: "index"
data_files:
- split: "noun"
path: "variants/index/noun.parquet"
- split: "verb"
path: "variants/index/verb.parquet"
- split: "adj"
path: "variants/index/adj.parquet"
- split: "adv"
path: "variants/index/adv.parquet"
- config_name: "exceptions"
data_files:
- split: "noun"
path: "variants/exceptions/noun.parquet"
- split: "verb"
path: "variants/exceptions/verb.parquet"
- split: "adj"
path: "variants/exceptions/adj.parquet"
- split: "adv"
path: "variants/exceptions/adv.parquet"
---

# WordNet 3.0

## Overview
- dataset_id: wordnet
- license: WordNet 3.0 license (Princeton)
- homepage: https://wordnet.princeton.edu

## Dataset Structure Guide
This dataset publishes multiple Hub configs so researchers can choose the right view for their workflow.

### What Configs Mean
A config is a named dataset view. It selects which files are loaded and which split names are exposed.

### Available Configs
- aggregate: Default aggregate view across available variant paths.
- splits: noun, verb, adj, adv
- data: Synset data files with glosses, words, pointers, and verb frames.
- splits: noun, verb, adj, adv
- file prefix: variants/data/
- index: Index files with lemma metadata, pointer symbol inventory, and synset offsets.
- splits: noun, verb, adj, adv
- file prefix: variants/index/
- exceptions: Exception files mapping inflected forms to base forms.
- splits: noun, verb, adj, adv
- file prefix: variants/exceptions/

### Quick Start (datasets library)
Use the config explicitly to avoid ambiguity and to get the expected split layout:

```python
from datasets import load_dataset

# Replace with your Hub repo id, e.g. '<namespace>/wordnet'
dataset = load_dataset('<namespace>/wordnet', 'aggregate')
print(dataset)
```

### Split Semantics
These splits may represent domain structure (for example part-of-speech groups) rather than train/validation/test partitions.
If you need ML evaluation splits, create an explicit derived split plan in downstream preprocessing.

## Summary
WordNet is published as lexical-resource views rather than benchmark train/validation/test tasks.

## Intended Use
- Lexical semantic analysis and symbolic NLP workflows.
- Knowledge graph and relation-structure exploration.
- Input resource for downstream task-specific data construction.

## Schema Notes
- The same part-of-speech split names (noun, verb, adj, adv) can appear across multiple configs.
- The aggregate config combines variant outputs while preserving split names.

## Evaluation Protocol
- For ML benchmarks, define explicit train/validation/test derivations from the lexical splits.
- Record any filtering or balancing logic to keep experiments reproducible.

## Limitations
- Not a native benchmark dataset with pre-defined evaluation partitions.
- Coverage and lexical conventions follow WordNet source files.

## Config Selection Guidance
- Use the data config for synset-level lexical structure.
- Use the index config for lemma and offset lookup behavior.
- Use the exceptions config for inflection-to-base mappings.
- Use aggregate when you want one entry point and can filter by variant in downstream code.

## Standardization
- split_aliases: {"dev": "validation", "val": "validation", "valid": "validation"}
- rename_map: {}
- selected_columns: []

## Exported Artifacts
- noun.parquet
- verb.parquet
- adj.parquet
- adv.parquet
- data
- WORDNET_FORMAT.md
- views/synsets.parquet
- views/words.parquet
- views/pointers.parquet
- views/frames.parquet

## Provided
- context: {"repository_navigation": [{"active": false, "description": "Synset data files with glosses, words, pointers, and verb frames.", "path_in_repo": "variants/data", "splits": ["noun", "verb", "adj", "adv"], "variant": "data"}, {"active": false, "description": "Index files with lemma metadata, pointer symbol inventory, and synset offsets.", "path_in_repo": "variants/index", "splits": ["noun", "verb", "adj", "adv"], "variant": "index"}, {"active": false, "description": "Exception files mapping inflected forms to base forms.", "path_in_repo": "variants/exceptions", "splits": ["noun", "verb", "adj", "adv"], "variant": "exceptions"}], "schema": ["text", "is_record", "offset", "synset_offset", "lex_filenum", "ss_type", "w_cnt", "word_count", "words", "lemmas", "p_cnt", "pointer_count", "pointers", "frames", "gloss", "parse_error", "lemma", "lemma_text", "pos", "synset_cnt", "ptr_symbols", "sense_cnt", "tagsense_cnt", "synset_offsets", "inflected_form", "inflected_form_text", "base_forms", "base_forms_text", "variant"], "scope": "aggregate", "shared_description": "WordNet publishes an aggregate lexical view that keeps the noun, verb, adjective, and adverb splits intact while combining data, index, and exception rows under an explicit variant column.", "splits": ["noun", "verb", "adj", "adv"], "viewer_config_name": "aggregate"}

## Repository Navigation
- data: variants/data | Synset data files with glosses, words, pointers, and verb frames.
- index: variants/index | Index files with lemma metadata, pointer symbol inventory, and synset offsets.
- exceptions: variants/exceptions | Exception files mapping inflected forms to base forms.

## Citation

Miller (1995), WordNet: A Lexical Database for English.

---
数据集展示名：词网（WordNet）3.0
许可证：其他
标签：
- 词典
- 词网（WordNet）
- 自然语言处理（NLP）
- 数据集
- 表格型数据
配置项：
- 配置名称："aggregate"
默认启用：true
数据文件：
- 划分："noun"（名词）
文件路径："noun.parquet"
- 划分："verb"（动词）
文件路径："verb.parquet"
- 划分："adj"（形容词）
文件路径："adj.parquet"
- 划分："adv"（副词）
文件路径："adv.parquet"
- 配置名称："data"
数据文件：
- 划分："noun"（名词）
文件路径："variants/data/noun.parquet"
- 划分："verb"（动词）
文件路径："variants/data/verb.parquet"
- 划分："adj"（形容词）
文件路径："variants/data/adj.parquet"
- 划分："adv"（副词）
文件路径："variants/data/adv.parquet"
- 配置名称："index"
数据文件：
- 划分："noun"（名词）
文件路径："variants/index/noun.parquet"
- 划分："verb"（动词）
文件路径："variants/index/verb.parquet"
- 划分："adj"（形容词）
文件路径："variants/index/adj.parquet"
- 划分："adv"（副词）
文件路径："variants/index/adv.parquet"
- 配置名称："exceptions"
数据文件：
- 划分："noun"（名词）
文件路径："variants/exceptions/noun.parquet"
- 划分："verb"（动词）
文件路径："variants/exceptions/verb.parquet"
- 划分："adj"（形容词）
文件路径："variants/exceptions/adj.parquet"
- 划分："adv"（副词）
文件路径："variants/exceptions/adv.parquet"
---

# 词网（WordNet）3.0

## 概览
- 数据集ID：wordnet
- 许可证：词网（WordNet）3.0许可证（普林斯顿大学）
- 官方主页：https://wordnet.princeton.edu

## 数据集结构指南
本数据集提供多种Hub配置项，便于研究人员根据工作流选择合适的视图。

### 配置项含义说明
配置项即命名的数据集视图，用于选择需加载的文件与暴露的划分名称。

### 可用配置项
- aggregate：默认的聚合视图，整合所有变体路径的数据集。
- 划分：名词、动词、形容词、副词
- data：包含释义、词汇、语义指针与动词框架的同义词集（synset）数据文件。
- 划分：名词、动词、形容词、副词
- 文件前缀：variants/data/
- index：包含词元（lemma）元数据、语义指针符号清单与同义词集偏移量的索引文件。
- 划分：名词、动词、形容词、副词
- 文件前缀：variants/index/
- exceptions：用于将屈折形式映射至基础形式的异常文件。
- 划分：名词、动词、形容词、副词
- 文件前缀：variants/exceptions/

### 快速上手（使用datasets库）
明确指定配置项可避免歧义并获得预期的划分布局：

python
from datasets import load_dataset

# 替换为你的Hub仓库ID，例如 '<namespace>/wordnet'
dataset = load_dataset('<namespace>/wordnet', 'aggregate')
print(dataset)

### 划分语义说明
此处的划分通常代表领域结构（如词性分组），而非训练/验证/测试集分区。若需用于机器学习评估，请在下游预处理阶段自行创建明确的派生划分方案。

## 数据集概述
词网（WordNet）以词汇资源视图的形式发布，而非基准测试用的训练/验证/测试任务集。

## 预期用途
- 词汇语义分析与符号式自然语言处理工作流
- 知识图谱与关系结构探索
- 下游任务专属数据构建的输入资源

## 模式说明
- 相同的词性划分名称（名词、动词、形容词、副词）可在多个配置项中出现。
- 聚合配置会整合变体输出，并保留划分名称。

## 评估协议
- 对于机器学习基准测试，请从词汇划分中自行定义明确的训练/验证/测试派生方案。
- 记录所有过滤或均衡逻辑，以确保实验可复现。

## 局限性说明
- 并非原生带有预定义评估分区的基准数据集。
- 覆盖范围与词汇约定遵循词网（WordNet）源文件标准。

## 配置项选择指南
- 如需进行同义词集级别的词汇结构分析，请使用data配置。
- 如需进行词元与偏移量查找，请使用index配置。
- 如需使用屈折形式到基础形式的映射，请使用exceptions配置。
- 如需单一入口点并可在下游代码中按变体进行过滤，请使用aggregate配置。

## 标准化规范
- 划分别名：{"dev": "validation", "val": "validation", "valid": "validation"}
- 重命名映射：{}
- 选定列：[]

## 导出产物
- noun.parquet
- verb.parquet
- adj.parquet
- adv.parquet
- data
- WORDNET_FORMAT.md
- views/synsets.parquet
- views/words.parquet
- views/pointers.parquet
- views/frames.parquet

## 提供内容
- 上下文：{
"仓库导航": [
{
"激活状态": false,
"描述": "包含释义、词汇、语义指针与动词框架的同义词集数据文件。",
"仓库内路径": "variants/data",
"划分": ["名词", "动词", "形容词", "副词"],
"变体": "data"
},
{
"激活状态": false,
"描述": "包含词元元数据、语义指针符号清单与同义词集偏移量的索引文件。",
"仓库内路径": "variants/index",
"划分": ["名词", "动词", "形容词", "副词"],
"变体": "index"
},
{
"激活状态": false,
"描述": "用于将屈折形式映射至基础形式的异常文件。",
"仓库内路径": "variants/exceptions",
"划分": ["名词", "动词", "形容词", "副词"],
"变体": "exceptions"
}
],
"模式字段": ["文本", "是否为记录", "偏移量", "同义词集偏移量", "词汇文件编号", "同义词集类型", "词汇计数", "词元计数", "词汇列表", "词元列表", "指针计数", "指针数量", "指针列表", "框架列表", "释义", "解析错误", "词元", "词元文本", "词性", "同义词集数量", "指针符号列表", "义项数量", "词标记计数", "同义词集偏移量列表", "屈折形式", "屈折形式文本", "基础形式列表", "基础形式文本", "变体"],
"作用域": "聚合",
"共享描述": "词网（WordNet）发布聚合词汇视图，保留名词、动词、形容词与副词划分，同时通过显式变体列整合数据、索引与异常文件行。",
"划分": ["名词", "动词", "形容词", "副词"],
"查看器配置项名称": "aggregate"
}

## 仓库导航
- 数据：variants/data | 包含释义、词汇、语义指针与动词框架的同义词集数据文件。
- 索引：variants/index | 包含词元元数据、语义指针符号清单与同义词集偏移量的索引文件。
- 异常文件：variants/exceptions | 用于将屈折形式映射至基础形式的异常文件。

## 引用文献
米勒（1995），《词网：面向英语的词汇数据库》。