README.md

---
license: cc-by-sa-3.0
task_categories:
- token-classification
language:
- en
tags:
- wikidata
- wikipedia
- named-entity-linking
pretty_name: AIDA CoNLL-YAGO Wikidata
size_categories:
- 10K<n<100K
---

# Dataset Card for AIDA CoNLL-YAGO Wikidata

## Table of Contents
- [Dataset Description](#dataset-description)
  - [Dataset Summary](#dataset-summary)
  - [Supported Tasks](#supported-tasks)
  - [Languages](#languages)
- [Dataset Structure](#dataset-structure)
  - [Data Instances](#data-instances)
  - [Data Fields](#data-fields)
  - [Data Splits](#data-splits)
- [Additional Information](#additional-information)
  - [Licensing Information](#licensing-information)

## Dataset Description

- **Repository:** [AIDA CoNLL-YAGO Wikidata repository](https://github.com/cyanic-selkie/aida-conll-yago-wikidata)

### Dataset Summary

The AIDA CoNLL-YAGO Wikidata dataset is the same as the original [AIDA CoNLL-YAGO](https://www.mpi-inf.mpg.de/departments/databases-and-information-systems/research/ambiverse-nlu/aida/downloads) dataset, but with Wikidata QIDs instead of Wikipedia titles as entity identifiers. They are automatically generated (with a few manual corrections) from Wikidata and Wikipedia dumps (March 1, 2023).

The code for generating the dataset can be found [here](https://github.com/cyanic-selkie/aida-conll-yago-wikidata).

### Supported Tasks

- `named-entity-linking`: The dataset can be used to train a model for Named Entity Linking.

### Languages

The text in the dataset is in English. The associated BCP-47 code is `en`.

## Dataset Structure

### Data Instances

A typical data point represents a sentence in a document (news article).

The `text` field contains the original text in an NFC normalized, UTF-8 encoded string.

The `entities` field contains a list of entities, each represented by a struct with the inclusive starting byte `start` field, exclusive ending byte `end` field, and a nullable `qid` field.

Additionally, each sentence has `document_id` and `sentence_index` fields to be able to determine the sentence's context.

Data points belonging to the same article appear next to each other in the dataset and are sorted in an ascending fashion by the `sentence_index` field.

An example from the AIDA CoNLL-YAGO Wikidata test set looks as follows:

```
{
  "uuid": "f647d183-ed81-4513-b607-2c9cbc98930a"
  "document_id": 1393,
  "sentence_index": 9,
  "text": "He guided Ireland to two successive World Cup finals tournaments and to the 1988 European championship finals in Germany , after the Irish beat a well-fancied England team 1-0 in their group qualifier .",
  "entities": [
    {
      "start": 10,
      "end": 17,
      "qid": 163547
    },
    {
      "start": 36,
      "end": 45,
      "qid": 19317
    },
    {
      "start": 81,
      "end": 89,
      "qid": 46
    },
    {
      "start": 113,
      "end": 120,
      "qid": 183
    },
    {
      "start": 133,
      "end": 138,
      "qid": 163547
    },
    {
      "start": 159,
      "end": 166,
      "qid": 47762
    }
  ]
}
```

### Data Fields

- `uuid`: a UTF-8 encoded string representing a v4 UUID that uniquely identifies the example
- `document_id`: an integer that uniquely identifies the document this sentence belongs to
- `sentence_index`: an integer that uniquely identifies the position of the sentence in its original document
- `text`: an NFC normalized, UTF-8 encoded string representing the sentence
- `entities`: a list of structs representing entities, each entity has:
  - `start`: an integer representing the inclusive starting UTF-8 code point of the entity
  - `end`: an integer representing the exclusive ending UTF-8 code point of the entity
  - `qid`: an integer representing the Wikidata QID this entity refers to; it can be null if the entity didn't exist in Wikidata at the time of the creation of the original dataset

### Data Splits

The data is split into training, validation and test sets; all of the sentences belonging to an article are in the same split. The final split sizes are as follows:

|                                               | Train     | Validation  | Test    |
| :-----                                        | :------:  | :-----:     | :----:  |
| AIDA CoNLL-YAGO Wikidata - documents          | 946       | 216         | 231     |
| AIDA CoNLL-YAGO Wikidata - sentences          | 13,938    | 3,225       | 3,421   |
| AIDA CoNLL-YAGO Wikidata - entities           | 23,374    | 5,912       | 5,608   |
| AIDA CoNLL-YAGO Wikidata - entities with QIDs | 18,540    | 4,791       | 4,481   |

## Additional Information

### Licensing Information

The licensing status of the dataset is the same as the licensing status of the original [AIDA CoNLL-YAGO](https://www.mpi-inf.mpg.de/departments/databases-and-information-systems/research/ambiverse-nlu/aida/downloads) dataset which is under a [Creative Commons Attribution-ShareAlike 3.0 Unported License](http://creativecommons.org/licenses/by-sa/3.0/deed.en_US).