Regular expressions for text cleaning in NLP - Model Pipeline Trace

Model Pipeline - Regular expressions for text cleaning

This pipeline shows how raw text data is cleaned using regular expressions before being used in a machine learning model. Cleaning removes unwanted characters and formats text for better learning.

Data Flow - 5 Stages

1Raw Text Input

1000 rows x 1 column→Original text data with noise like punctuation, numbers, and mixed cases→1000 rows x 1 column

"Hello!!! This is a test, number 123."

↓

2Lowercasing

1000 rows x 1 column→Convert all text to lowercase→1000 rows x 1 column

"hello!!! this is a test, number 123."

↓

3Remove Punctuation

1000 rows x 1 column→Use regex to remove punctuation marks→1000 rows x 1 column

"hello this is a test number 123"

↓

4Remove Numbers

1000 rows x 1 column→Use regex to remove digits→1000 rows x 1 column

"hello this is a test number "

↓

5Remove Extra Spaces

1000 rows x 1 column→Use regex to replace multiple spaces with a single space→1000 rows x 1 column

"hello this is a test number"

Training Trace - Epoch by Epoch

Loss
1.0 |          *
0.8 |        *  
0.6 |      *    
0.4 |    *      
0.2 |  *        
0.0 +-----------
      1 2 3 4 5
       Epochs

Epoch	Loss ↓	Accuracy ↑	Observation
1	0.85	0.60	Initial training with raw text features, loss high due to noise
2	0.65	0.72	After cleaning text with regex, model starts learning better
3	0.50	0.80	Loss decreases and accuracy improves as text is cleaner
4	0.40	0.85	Model converges with clean text input
5	0.35	0.88	Final epoch shows stable loss and high accuracy

Prediction Trace - 5 Layers

Layer 1: Input Raw Text

Layer 2: Lowercasing

Layer 3: Remove Punctuation

Layer 4: Remove Numbers

Layer 5: Remove Extra Spaces

Model Quiz - 3 Questions

What does the regex step 'Remove Punctuation' do to the text?

A Deletes symbols like commas and exclamation marks

B Changes all letters to uppercase

C Removes all spaces between words

D Adds numbers to the text

Key Insight

Cleaning text with regular expressions removes noise like punctuation and numbers, making the data easier for the model to learn from. This leads to lower loss and higher accuracy during training.