Implement structure substitute
- Dominant language
- Python
- Stars
- 253
- Forks
- 59
- PR merge metrics
- No merged PRs in 30d
Description
**Is your feature request related to a problem? Please describe.**
Structure substitution is a general family of methods in data augmentation. https://home.ttic.edu/~freda/paper/shi2021substructure.pdf
Behind the scene, the method focuses on replacing sub-structures like phrases or trees from a sentence. This fits very well into Forte's data augmentation framework, where the systems allow one to replace content but keeping the original annotation.
**Describe the solution you'd like**
1. Implement structure substitution as a `ReplacementOp`.
2. A pipeline needs to be run to collect the sub-structure statistics first.
3. Structured constraints (as mentioned in the paper) can be considered as the configurations for the processor
**Describe alternatives you've considered**
A clear and concise description of any alternative solutions or features you've considered.
**Additional context**
Add any other context or screenshots about the feature request here.
Contributor guide
Research direction
Start by reading the linked structure-substitution paper and Forte's existing ReplacementOp implementations. Trace how augmentation pipelines collect statistics and preserve annotations, then determine how structured constraints should be configured. Done means a working structure-substitution operation integrated with the pipeline and covered by appropriate tests; the issue names no files or tests.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 25/100