---
dataset_info:
  features:
  - name: index
    dtype: int64
  - name: label
    dtype: int64
  - name: sentence
    dtype: string
  - name: pos
    dtype: string
  - name: v_index
    dtype: int64
  splits:
  - name: train
    num_bytes: 6970850
    num_examples: 37370
  download_size: 4354865
  dataset_size: 6970850
license: cc-by-2.0
---

# TroFi_Metaphor

## Dataset Description

- **Paper:** [A Clustering Approach for the Nearly Unsupervised Recognition of Nonliteral Language](https://aclanthology.org/E06-1042.pdf)

## Dataset Summary

The TroFi (Trope Finder) dataset is an unsupervised collection of data specifically designed to classify verbs into either literal or nonliteral categories. This dataset is composed of three primary sets. Firstly, the Target Set, which includes sentences featuring the verbs to be classified. These sentences are extracted from the '88-'89 Wall Street Journal (WSJ) Corpus and tagged using specific tagging systems, namely Ratnaparkhi's tagger and Bangalore & Joshi's SuperTagger. Secondly, there's the Literal Feedback Set, which consists of sentences from the WSJ Corpus that contain seed words derived from WordNet, used to provide a literal context. Finally, the Nonliteral Feedback Set comprises sentences from the WSJ that contain seed words drawn from known databases of metaphors, idioms, and expressions. These include Wayne Magnuson's English Idioms Sayings & Slang and George Lakoff’s Conceptual Metaphor List. The TroFi dataset employs an automated method to minimize the potential negative impact of unverified "literalness" in the feedback sets and to manage instances where nonliteral sets are sparse. The primary goal of the TroFi dataset is to recognize instances of nonliteral language that may not be fully covered by existing databases, thereby enhancing our ability to determine when an expression is being used nonliterally.
For the details of this dataset, we refer you to the original [paper](https://aclanthology.org/E06-1042.pdf).

Metadata in Creative Language Toolkit ([CLTK](https://github.com/liyucheng09/cltk))
- CL Type: Metaphor
- Task Type: detection
- Size: 37k
- Created time: 2006

### Citation Information

If you find this dataset helpful, please cite:

```
@inproceedings{Birke2006ACA,
  title={A Clustering Approach for Nearly Unsupervised Recognition of Nonliteral Language},
  author={Julia Birke and Anoop Sarkar},
  booktitle={Conference of the European Chapter of the Association for Computational Linguistics},
  year={2006}
}
```

### Contributions

If you have any queries, please open an issue or direct your queries to [mail](mailto:yucheng.li@surrey.ac.uk).