File size: 14,769 Bytes
f96b5f6 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 |
---
license: cc-by-sa-4.0
task_categories:
- text-retrieval
language:
- nl
pretty_name: BEIR-NL
size_categories:
- 100K<n<1M
configs:
- config_name: default
data_files:
- split: queries
path: queries.jsonl
- split: corpus
path: corpus.jsonl
---
# Dataset Card for BEIR-NL Benchmark
## Table of Contents
- [Dataset Description](#dataset-description)
- [Dataset Summary](#dataset-summary)
- [Languages](#languages)
- [Dataset Structure](#dataset-structure)
- [Data Instances](#data-instances)
- [Data Fields](#data-fields)
- [Data Splits](#data-splits)
- [Dataset Creation](#dataset-creation)
- [Curation Rationale](#curation-rationale)
- [Source Data](#source-data)
- [Annotations](#annotations)
- [Considerations for Using the Data](#considerations-for-using-the-data)
- [Other Known Limitations](#other-known-limitations)
- [Additional Information](#additional-information)
- [Licensing Information](#licensing-information)
- [Citation Information](#citation-information)
## Dataset Description
- **Paper:** [BEIR-NL: Zero-shot Information Retrieval Benchmark for the Dutch Language](https://arxiv.org/abs/2412.08329)
### Dataset Summary
BEIR-NL is a Dutch-translated version of the BEIR benchmark, a diverse and heterogeneous collection of datasets covering various domains from biomedical and financial texts to general web content.
BEIR-NL contains the following tasks:
- Fact-checking: [FEVER](http://fever.ai), [Climate-FEVER](http://climatefever.ai), [SciFact](https://github.com/allenai/scifact)
- Question-Answering: [NQ](https://ai.google.com/research/NaturalQuestions), [HotpotQA](https://hotpotqa.github.io), [FiQA-2018](https://sites.google.com/view/fiqa/)
- Bio-Medical IR: [TREC-COVID](https://ir.nist.gov/covidSubmit/index.html), [NFCorpus](https://www.cl.uni-heidelberg.de/statnlpgroup/nfcorpus/)
- Argument Retrieval: [Touche-2020](https://webis.de/events/touche-20/shared-task-1.html), [ArguAna](tp://argumentation.bplaced.net/arguana/data)
- Duplicate Question Retrieval: [Quora](https://www.quora.com/q/quoradata/First-Quora-Dataset-Release-Question-Pairs), [CqaDupstack](http://nlp.cis.unimelb.edu.au/resources/cqadupstack/)
- Citation-Prediction: [SCIDOCS](https://allenai.org/data/scidocs)
- Entity Retrieval: [DBPedia](https://github.com/iai-group/DBpedia-Entity/)
### Languages
Dutch
## Dataset Structure
BEIR-NL adheres to the structure of the original BEIR benchmark. All BEIR datasets must contain a corpus, queries and qrels (relevance judgments file). Qrels can be found in the BEIR repository on [GitHub](https://github.com/beir-cellar/beir) or [Hugging Face](https://huggingface.co/BeIR). They format:
- `corpus` file: a `.jsonl` file (jsonlines) that contains a list of dictionaries, each with three fields `_id` with unique document identifier, `title` with document title (optional) and `text` with document paragraph or passage. For example: `{"_id": "doc1", "title": "Albert Einstein", "text": "Albert Einstein was een in Duitsland geboren..."}`
- `queries` file: a `.jsonl` file (jsonlines) that contains a list of dictionaries, each with two fields `_id` with unique query identifier and `text` with query text. For example: `{"_id": "q1", "text": "Wie ontwikkelde de massa-energie-equivalentieformule?"}`
- `qrels` file: a `.tsv` file (tab-seperated) that contains three columns, i.e. the `query-id`, `corpus-id` and `score` in this order. Keep 1st row as header. For example: `q1 doc1 1`
### Data Instances
A high level example of any beir dataset:
```python
corpus = {
"doc1": {
"title": "Albert Einstein",
"text": (
"Albert Einstein was een in Duitsland geboren theoretisch natuurkundige die de relativiteitstheorie ontwikkelde, "
"een van de twee pijlers van de moderne natuurkunde (samen met de kwantummechanica). Zijn werk staat ook bekend "
"om zijn invloed op de wetenschapfilosofie. Hij is bij het grote publiek vooral bekend vanwege zijn massa-energie- "
"equivalentieformule E = mc^2, die 's werelds beroemdste vergelijking' wordt genoemd. Hij ontving in 1921 de Nobelprijs "
"voor de Natuurkunde 'voor zijn verdiensten aan de theoretische natuurkunde, en in het bijzonder voor zijn ontdekking "
"van de wet van het foto-elektrisch effect', een cruciale stap in de ontwikkeling van de kwantumtheorie."
),
},
"doc2": {
"title": "",
"text": (
"Tarwebier is een bovengistend bier dat wordt gebrouwen met een groot aandeel tarwe ten opzichte van de hoeveelheid "
"gemoute gerst. De twee belangrijkste soorten zijn Duits Weißbier en Belgisch witbier; andere soorten zijn onder andere "
"Lambiek (gemaakt met wilde gist), Berliner Weisse (een troebel, zuur bier) en Gose (een zuur, zout bier)."
),
},
}
queries = {
"q1": "Wie ontwikkelde de massa-energie-equivalentieformule?",
"q2": "Welk bier wordt gebrouwen met een groot aandeel tarwe?"
}
qrels = {
"q1": {"doc1": 1},
"q2": {"doc2": 1},
}
```
### Data Fields
Examples from all configurations have the following features:
#### Corpus
- `corpus`: a `dict` feature representing the document title and passage text, made up of:
- `_id`: a `string` feature representing the unique document id
- `title`: a `string` feature, denoting the title of the document.
- `text`: a `string` feature, denoting the text of the document.
#### Queries
- `queries`: a `dict` feature representing the query, made up of:
- `_id`: a `string` feature representing the unique query id
- `text`: a `string` feature, denoting the text of the query.
#### Qrels
- `qrels`: a `dict` feature representing the query document relevance judgements, made up of:
- `_id`: a `string` feature representing the query id
- `_id`: a `string` feature, denoting the document id.
- `score`: a `int32` feature, denoting the relevance judgement between query and document.
### Data Splits
| Dataset | Website| BEIR-Name | Type | Queries | Corpus | Rel D/Q | BEIR | BEIR-NL |
| -------- | -----| ---------| --------- | ----------- | ---------| ---------|:-----------------------------------------------------------------------------------------------:|:-------------------------------------------------------------------------:|
| TREC-COVID | [Homepage](https://ir.nist.gov/covidSubmit/index.html)| ``trec-covid``| ``test``| 50| 171K| 493.5 | [Link](https://public.ukp.informatik.tu-darmstadt.de/thakur/BEIR/datasets/trec-covid.zip) | [Link](https://huggingface.co/datasets/clips/beir-nl-trec-covid) |
| NFCorpus | [Homepage](https://www.cl.uni-heidelberg.de/statnlpgroup/nfcorpus/) | ``nfcorpus`` | ``train``<br>``dev``<br>``test``| 323 | 3.6K | 38.2 | [Link](https://public.ukp.informatik.tu-darmstadt.de/thakur/BEIR/datasets/nfcorpus.zip) | [Link]( [Link](https://huggingface.co/datasets/clips/beir-nl-trec-covid)) |
| NQ | [Homepage](https://ai.google.com/research/NaturalQuestions) | ``nq``| ``train``<br>``test``| 3,452 | 2.68M | 1.2 | [Link](https://public.ukp.informatik.tu-darmstadt.de/thakur/BEIR/datasets/nq.zip) | [Link](https://huggingface.co/datasets/clips/beir-nl-nq) |
| HotpotQA | [Homepage](https://hotpotqa.github.io) | ``hotpotqa``| ``train``<br>``dev``<br>``test``| 7,405 | 5.23M | 2.0 | [Link](https://public.ukp.informatik.tu-darmstadt.de/thakur/BEIR/datasets/hotpotqa.zip) | [Link](https://huggingface.co/datasets/clips/beir-nl-hotpotqa) |
| FiQA-2018 | [Homepage](https://sites.google.com/view/fiqa/) | ``fiqa`` | ``train``<br>``dev``<br>``test``| 648 | 57K | 2.6 | [Link](https://public.ukp.informatik.tu-darmstadt.de/thakur/BEIR/datasets/fiqa.zip) | [Link](https://huggingface.co/datasets/clips/beir-nl-fiqa) |
| ArguAna | [Homepage](http://argumentation.bplaced.net/arguana/data) | ``arguana``| ``test`` | 1,406 | 8.67K | 1.0 | [Link](https://public.ukp.informatik.tu-darmstadt.de/thakur/BEIR/datasets/arguana.zip) | [Link](https://huggingface.co/datasets/clips/beir-nl-arguana) |
| Touche-2020| [Homepage](https://webis.de/events/touche-20/shared-task-1.html) | ``webis-touche2020``| ``test``| 49 | 382K | 19.0 | [Link](https://public.ukp.informatik.tu-darmstadt.de/thakur/BEIR/datasets/webis-touche2020.zip) | [Link](https://huggingface.co/datasets/clips/beir-nl-webis-touche2020) |
| CQADupstack| [Homepage](http://nlp.cis.unimelb.edu.au/resources/cqadupstack/) | ``cqadupstack``| ``test``| 13,145 | 457K | 1.4 | [Link](https://public.ukp.informatik.tu-darmstadt.de/thakur/BEIR/datasets/cqadupstack.zip) | [Link](https://huggingface.co/datasets/clips/beir-nl-cqadupstack) |
| Quora| [Homepage](https://www.quora.com/q/quoradata/First-Quora-Dataset-Release-Question-Pairs) | ``quora``| ``dev``<br>``test``| 10,000 | 523K | 1.6 | [Link](https://public.ukp.informatik.tu-darmstadt.de/thakur/BEIR/datasets/quora.zip) | [Link](https://huggingface.co/datasets/clips/beir-nl-quora) |
| DBPedia | [Homepage](https://github.com/iai-group/DBpedia-Entity/) | ``dbpedia-entity``| ``dev``<br>``test``| 400 | 4.63M | 38.2 | [Link](https://public.ukp.informatik.tu-darmstadt.de/thakur/BEIR/datasets/dbpedia-entity.zip) | [Link](https://huggingface.co/datasets/clips/beir-nl-dbpedia-entity) |
| SCIDOCS| [Homepage](https://allenai.org/data/scidocs) | ``scidocs``| ``test``| 1,000 | 25K | 4.9 | [Link](https://public.ukp.informatik.tu-darmstadt.de/thakur/BEIR/datasets/scidocs.zip) | [Link](https://huggingface.co/datasets/clips/beir-nl-scidocs) |
| FEVER | [Homepage](http://fever.ai) | ``fever``| ``train``<br>``dev``<br>``test``| 6,666 | 5.42M | 1.2| [Link](https://public.ukp.informatik.tu-darmstadt.de/thakur/BEIR/datasets/fever.zip) | [Link](https://huggingface.co/datasets/clips/beir-nl-fever) |
| Climate-FEVER| [Homepage](http://climatefever.ai) | ``climate-fever``|``test``| 1,535 | 5.42M | 3.0 | [Link](https://public.ukp.informatik.tu-darmstadt.de/thakur/BEIR/datasets/climate-fever.zip) | [Link](https://huggingface.co/datasets/clips/beir-nl-climate-fever) |
| SciFact| [Homepage](https://github.com/allenai/scifact) | ``scifact``| ``train``<br>``test``| 300 | 5K | 1.1 | [Link](https://public.ukp.informatik.tu-darmstadt.de/thakur/BEIR/datasets/scifact.zip) | [Link](https://huggingface.co/datasets/clips/beir-nl-scifact) |
## Dataset Creation
### Curation Rationale
Zero-shot evaluation of information retrieval (IR) models is often performed using BEIR; a large and heterogeneous benchmark composed of multiple datasets, covering different retrieval tasks across various domains. Although BEIR has become a standard benchmark for the zero-shot setup, its exclusively English content reduces its utility for underrepresented languages in IR, including Dutch. To address this limitation and encourage the development of Dutch IR models, we introduce BEIR-NL by automatically translating the publicly accessible BEIR datasets into Dutch.
### Source Data
BEIR repository on [GitHub](https://github.com/beir-cellar/beir).
### Annotations
We prompted Gemini-1.5-flash to translate BEIR into Dutch. A small portion of translations were done using GPT-4o-mini and Google Translate, as Gemini declined to translate certain content and had occasional issues with tags in prompts.
## Considerations for Using the Data
### Other Known Limitations
**Not Native Dutch Resources.** While BEIR-NL provides a benchmark for evaluating IR models in Dutch, it relies on translations from the original BEIR, which is exclusively in English. This lack of native Dutch datasets limits the ability of BEIR-NL to fully represent and reflect the linguistic nuances and cultural context of the language, and therefore, the complexities of Dutch IR, especially in domain-specific contexts with local terminology and knowledge.
**Data Contamination.** Many modern IR models are trained on massive corpora that might include content from BEIR. This can result in inflated performances --as models might have already seen the relevant data during different phases of training-- raising concerns about the validity of zero-shot evaluations. Ensuring a truly zero-shot evaluation is a difficult challenge, as many IR models lack transparency regarding the exact composition of training corpora.
**Benchmark Validity Over Time.** BEIR has become a standard benchmark to evaluate the performance of IR models, attracting a large number of evaluations over time. This extensive usage introduces the risk of overfitting, as researchers might unintentionally train models tailored to perform well on BEIR rather than on broader IR tasks. In addition, advances in IR models and evaluation needs might outpace the benchmark, making it less representative and less relevant. As a result, the relevance and validity of BEIR as well as BEIR-NL may diminish over time.
## Additional Information
### Licensing Information
This subset (Touche-2020) of BEIR-NL is licensed under the [CC BY-NC-SA 4.0 license](https://creativecommons.org/licenses/by-nc-sa/4.0/).
### Citation Information
If you find BEIR-NL useful in your research, please consider citing it, as well as the original BEIR benchmark it is derived from:
```
@misc{banar2024beirnlzeroshotinformationretrieval,
title={BEIR-NL: Zero-shot Information Retrieval Benchmark for the Dutch Language},
author={Nikolay Banar and Ehsan Lotfi and Walter Daelemans},
year={2024},
eprint={2412.08329},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2412.08329},
}
@inproceedings{thakur2021beir,
title={{BEIR}: A Heterogeneous Benchmark for Zero-shot Evaluation of Information Retrieval Models},
author={Nandan Thakur and Nils Reimers and Andreas R{\"u}ckl{\'e} and Abhishek Srivastava and Iryna Gurevych},
booktitle={Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Track (Round 2)},
year={2021},
url={https://openreview.net/forum?id=wCu6T5xFjeJ}
}
```
|