Data Augmentation

Endpoint: POST /data-augmentation

Request Parameters

File Upload

• file (required): The file to be processed. Supported formats include CSV, Excel, etc.

Form Parameters

• method (optional, default: word_shuffling): The data augmentation method to apply. Supported values:
  • word_shuffling: Shuffles words in the text.
  • sentence_shuffling: Shuffles sentences in the text.
  • word_replacement: Replaces words using a BERT-based model.
  • syntax_tree_manipulation: Manipulates the syntax tree of the text.
  • random_word_insertion: Inserts random words into the text.
  • random_word_deletion: Deletes random words from the text.
• column (required): The column in the file to apply the augmentation to.
• num_augmented (optional, default: 1): The number of augmented rows to generate for each original row.

Processing Logic

1. Read the uploaded file: The file is converted into a Pandas DataFrame.
2. Validate the column: Checks if the specified column exists in the DataFrame.
3. Apply the selected augmentation method:
   • Word Shuffling: Shuffles the words in the text.
   • Sentence Shuffling: Shuffles the sentences in the text.
   • Word Replacement: Replaces words using a BERT-based model.
   • Syntax Tree Manipulation: Manipulates the syntax tree of the text.
   • Random Word Insertion: Inserts random words into the text.
   • Random Word Deletion: Deletes random words from the text.
4. Generate augmented rows: For each original row, the specified number of augmented rows is generated.
5. Return the modified file: The augmented file is provided for download with the filename appended with _augmented.

Example Request

curl -X POST "http://localhost:8000/data-augmentation" \
-F "file=@data.csv" \
-F "method=word_shuffling" \
-F "column=text" \
-F "num_augmented=2"