UiPath Document Understanding Data Labelling

Help Activities
, ,
1

I have used Data labelling with 25 documents. However to my dismay I found that in one fot he documents I have labelled wrong fields. If I relabel that partifular document then do I need to do the following ?

  1. Export the data
  2. Run the pipeline once again
1 Like
2

Let’s go one by one

Yes, it’s advisable to export the data again before relabeling a document. This is because the pipeline will retrain on all of the labeled data, including the relabeled document.

Exporting the data will ensure that the changes you made to the labels are incorporated into the dataset you use for your machine learning model or analysisIf you do not export the data, the pipeline will not be able to learn from the relabeled document.

Yes, you should run the pipeline once again after relabeling a document. This is because the pipeline needs to be retrained on the new data in order to produce accurate results.

If you are only relabeling a few documents, you may be able to get away with not exporting and importing the data. However, it is generally a good practice to export and import the data when relabeling documents. This is because it ensures that the labeling pipeline is trained on the most up-to-date data.

Hope this clarifies

Cheers @Ritaman_Baral

3

@Ritaman_Baral

Correct the labeling for the specific document where you identified the issue. Make sure you label the fields correctly this time.
If you’re using a data labeling tool or platform, make sure to save or update the labeling data for the corrected document. This might involve making changes within the tool’s interface to adjust the labels associated with that document.
Depending on your data labeling process, you may need to export the corrected labeling data for the specific document. Ensure that the updated data accurately reflects the corrections you made.
Once you have the corrected labeling data for the specific document, you may need to re-run the pipeline or process that uses this data. This is crucial to ensure that the corrections are incorporated into your downstream data processing tasks.

Cheers…!

1 Like
4

Thanks for your explanation.

1 Like
5

This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.