asyml / asyml/forte

Check Correctness of Copying Link and Group during Augmentation

Open
#712 0 comments 0 reactions 0 assignees View on GitHub
bug data_aug e/3
Dominant language
Python
Stars
253
Forks
59
PR merge metrics
No merged PRs in 30d

Description

**Describe the bug**
During augmentation, the newly created data pack copies the links and groups of the original data pack. We suspect that the user defined features which may be a part of the group may not be getting copied.

**To Reproduce**
Steps to reproduce the behavior:
This can be check in the tests of the `BaseOpProcessor`:
1) https://github.com/Pushkar-Bhuse/forte/blob/131f80fde277d9471332db670f9dd6b6ff64710d/tests/forte/processors/data_augment/algorithms/base_data_augmentation_op_test.py#L203
2) https://github.com/Pushkar-Bhuse/forte/blob/131f80fde277d9471332db670f9dd6b6ff64710d/tests/forte/processors/data_augment/algorithms/base_data_augmentation_op_test.py#L453

However, in these tests, the copying of user defined attributes of links and groups is not tested. This is the behavior that needs to be tested. If the above mentioned concern is true, then there need to be changes made in how links and groups are copied here:
https://github.com/asyml/forte/blob/92fe7280f7faddbf7b67a3bbc174f316caf24d62/forte/processors/data_augment/algorithms/base_data_augmentation_op.py#L125

Contributor guide

Open the contributing guide

Research direction

Start with tests/forte/processors/data_augment/algorithms/base_data_augmentation_op_test.py at the referenced BaseOpProcessor tests, then inspect forte/processors/data_augment/algorithms/base_data_augmentation_op.py around line 125. Add coverage for user-defined attributes on copied links and groups. Done means the tests verify those attributes are preserved, with implementation changes only if the tests expose a copying bug.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Clearly specified
Newbie friendliness
48/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.