What is Auto-labeling?
Auto labeling or pre-annotation allows enrichment, annotation, and labeling of datasets that are used for training AI models.
UBIAI has an auto-labeling feature that decreases human effort and saves time and money on data labeling.
The tool offers an option of auto-annotation of documents by using an ML model and dictionaries.
It easily allows auto-annotation of entities such as time, location, date, product, person, etc., after uploading the text from native files.
Auto-Labeling with Dictionary
Using UBIAI’s auto-labeling feature can be helpful in the following ways:
For auto-labeling and annotation of words with UBIAI, you can associate one or more dictionaries.
The dictionary will help in entity labeling and, ultimately, pre-annotation of data.
The dictionary file should contain every associated word with the corresponding entity type.
The tool will automatically annotate the text present in the dictionary and the corpus document.
Auto-Labeling with ML Models
UBIAI allows auto-labeling of the uploaded document(s) using a fine-tuned ML model on your annotation.
UBIAI offers the option fine-tune state-of-the-art transformer models.
After completing the training, UBIAI’s tool will evaluate the model performance.
For each entity, the F score, precision, and recall score will be displayed.
After model training, you have the option to export the ready-to-use model for your application.
Rule-based Auto-labeling
Using UBIAI’s auto-labeling feature can be helpful in the following ways:
For auto-labeling and annotation of words with UBIAI, you can associate one or more dictionaries.
The dictionary will help in entity labeling and, ultimately, pre-annotation of data.
The dictionary file should contain every associated word with the corresponding entity type.
Rule-based matching allows auto-labeling of documents by combining multiple rules.
It will enable instant auto-labeling of documents by using multiple pre-defined rules such as regular expressions, Part of Speech (POS), and patterns (email, number, phone number, etc.…).
Explore other Features
Proud to show how well
our users see us.
We keep shining all over! Named High Performer for
two season on a row!
Ihsan S.
Machine Learning Engineer
Jetson W.
Data Architect
Jasmeet K.
Data Scientist
What are you waiting for?
Automate your process!
