README.md

---
annotations_creators:
- expert-generated
- machine-generated
language:
- en
license: mit
multilinguality: monolingual
pretty_name: SemanticQA
size_categories:
- 1K<n<10K
source_datasets:
- original
tags:
- semantics
- idioms
- noun-compounds
- collocations
- multiword-expressions
- benchmark
- nlp
task_categories:
- text-classification
- text-generation
- question-answering
task_ids:
- multiple-choice-qa
- open-domain-qa
dataset_info:
- config_name: collocate_retrieval
  splits:
  - name: test
    num_examples: 306
- config_name: collocation_categorization
  splits:
  - name: test
    num_examples: 305
- config_name: collocation_extraction
  splits:
  - name: test
    num_examples: 305
- config_name: collocation_paraphrase
  splits:
  - name: test
    num_examples: 305
- config_name: idiom_detection
  splits:
  - name: test
    num_examples: 273
- config_name: idiom_extraction
  splits:
  - name: test
    num_examples: 447
- config_name: idiom_paraphrase
  splits:
  - name: test
    num_examples: 818
- config_name: noun_compound_compositionality
  splits:
  - name: test
    num_examples: 242
- config_name: noun_compound_compositionality_ft
  splits:
  - name: train
    num_examples: 121
  - name: test
    num_examples: 97
  - name: validation
    num_examples: 24
- config_name: noun_compound_extraction
  splits:
  - name: test
    num_examples: 720
- config_name: noun_compound_interpretation
  splits:
  - name: test
    num_examples: 110
- config_name: verbal_mwe_extraction
  splits:
  - name: test
    num_examples: 475
configs:
- config_name: collocate_retrieval
  data_files:
  - split: test
    path: data/collocate_retrieval/collocate_retrieval.json
- config_name: collocation_categorization
  data_files:
  - split: test
    path: data/collocation_categorization/collocation_categorization.json
- config_name: collocation_extraction
  data_files:
  - split: test
    path: data/collocation_extraction/collocation_extraction.json
- config_name: collocation_paraphrase
  data_files:
  - split: test
    path: data/collocation_paraphrase/collocation_paraphrase.json
- config_name: idiom_detection
  data_files:
  - split: test
    path: data/idiom_detection/idiom_detection.json
  default: true
- config_name: idiom_extraction
  data_files:
  - split: test
    path: data/idiom_extraction/idiom_extraction.json
- config_name: idiom_paraphrase
  data_files:
  - split: test
    path: data/idiom_paraphrase/idiom_paraphrase.json
- config_name: noun_compound_compositionality
  data_files:
  - split: test
    path: data/noun_compound_compositionality/noun_compound_compositionality.json
- config_name: noun_compound_compositionality_ft
  data_files:
  - split: train
    path: data/noun_compound_compositionality/noun_compound_compositionality_ft_train.json
  - split: test
    path: data/noun_compound_compositionality/noun_compound_compositionality_ft_test.json
  - split: validation
    path: data/noun_compound_compositionality/noun_compound_compositionality_ft_valid.json
- config_name: noun_compound_extraction
  data_files:
  - split: test
    path: data/noun_compound_extraction/noun_compound_extraction.json
- config_name: noun_compound_interpretation
  data_files:
  - split: test
    path: data/noun_compound_interpretation/noun_compound_interpretation.json
- config_name: verbal_mwe_extraction
  data_files:
  - split: test
    path: data/verbal_mwe_extraction/verbal_mwe_extraction.json
---

# SemanticQA

A comprehensive benchmark for evaluating language models on semantic phrase processing, from the paper [*Revisiting a Pain in the Neck: Semantic Phrase Processing Benchmark for Language Models*](https://arxiv.org/abs/2405.02861).

## Usage

```python
from datasets import load_dataset

# Load a specific subset
dataset = load_dataset("jacklanda/SemanticQA", "idiom_detection")

# Available configs:
#   collocate_retrieval, collocation_categorization, collocation_extraction,
#   collocation_paraphrase, idiom_detection, idiom_extraction, idiom_paraphrase,
#   noun_compound_compositionality, noun_compound_compositionality_ft,
#   noun_compound_extraction, noun_compound_interpretation, verbal_mwe_extraction
```

## Subsets

| Config | Task | Phrase Type | Size | Eval Metrics |
|--------|------|-------------|------|--------------|
| `collocate_retrieval` | Collocate Retrieval (CR) | Collocation | 306 | Exact Match |
| `collocation_categorization` | Collocation Categorization (LCC) | Collocation | 305 | Accuracy, F1 |
| `collocation_extraction` | Collocation Extraction (LCE) | Collocation | 305 | Exact Match |
| `collocation_paraphrase` | Collocation Interpretation (LCI) | Collocation | 305 | ROUGE-L, BERTScore, METEOR, BLEU |
| `idiom_detection` | Idiom Detection (IED) | Idiom | 273 | MCQ Accuracy |
| `idiom_extraction` | Idiom Extraction (IEE) | Idiom | 447 | Exact Match |
| `idiom_paraphrase` | Idiom Interpretation (IEI) | Idiom | 818 | ROUGE-L, BERTScore, METEOR, BLEU |
| `noun_compound_compositionality` | NC Compositionality (NCC) | Noun Compound | 242 | MCQ Accuracy |
| `noun_compound_compositionality_ft` | NCC Fine-tuning splits | Noun Compound | 242 | — |
| `noun_compound_extraction` | NC Extraction (NCE) | Noun Compound | 720 | Exact Match |
| `noun_compound_interpretation` | NC Interpretation (NCI) | Noun Compound | 110 | ROUGE-L, BERTScore, METEOR, BLEU |
| `verbal_mwe_extraction` | VMWE Extraction | Verbal MWE | 475 | Exact Match |

## Citation

```bibtex
@article{liu2026revisiting,
    title={Revisiting a Pain in the Neck: A Semantic Reasoning Benchmark for Language Models},
    author={Liu, Yang and Li, Hongming and Qin, Melissa Xiaohui and Liu, Qiankun and Huang, Chao},
    journal={arXiv preprint arXiv:2604.16593},
    year={2026}
}
```

```bibtex
@article{liu2024revisiting,
  title={Revisiting a Pain in the Neck: Semantic Phrase Processing Benchmark for Language Models},
  author={Liu, Yang and Qin, Melissa Xiaohui and Li, Hongming and Huang, Chao},
  journal={arXiv preprint arXiv:2405.02861},
  year={2024}
}
```

## License

MIT