- Title
- Expressing metaphorically, writing creatively: Metaphor identification for creativity assessment
- Creator
- Zhang, Dongyu; Zhang, Minghao; Peng, Ciyuan; Xia, Feng
- Date
- 2022
- Type
- Text; Conference proceedings
- Identifier
- http://researchonline.federation.edu.au/vital/access/HandleResolver/1959.17/188890
- Identifier
- vital:17357
- Identifier
-
https://doi.org/10.1145/3487553.3524935
- Identifier
- ISBN:978-1-4503-9130-6
- Abstract
- Metaphor, which can implicitly express profound meanings and emotions, is a unique writing technique frequently used in human language. In writing, meaningful metaphorical expressions can enhance the literariness and creativity of texts. Therefore, the usage of metaphor is a significant impact factor when assessing the creativity and literariness of writing. However, little to no automatic writing assessment system considers metaphorical expressions when giving the score of creativity. For improving the accuracy of automatic writing assessment, this paper proposes a novel creativity assessment model that imports a token-level metaphor identification method to extract metaphors as the indicators for creativity scoring. The experimental results show that our model can accurately assess the creativity of different texts with precise metaphor identification. To the best of our knowledge, we are the first to apply automatic metaphor identification to assess writing creativity. Moreover, identifying features (e.g., metaphors) that influence writing creativity using computational approaches can offer fair and reliable assessment methods for educational settings.
- Publisher
- ACM
- Relation
- WWW '22: Companion Proceedings of the Web Conference , Virtual event , April 2022 p. 1198-
- Rights
- All metadata describing materials held in, or linked to, the repository is freely available under a CC0 licence
- Rights
- © 2022 Association for Computing Machinery
- Rights
- Open Access
- Full Text
- Reviewed
- Hits: 395
- Visitors: 363
- Downloads: 10
| Thumbnail | File | Description | Size | Format |
|---|