NER with spaCy in NLP - Model Pipeline Trace

Model Pipeline - NER with spaCy

This pipeline detects named entities like names, places, and dates in text using spaCy's NER model. It reads raw text, processes it to find entities, and outputs labeled text with entity types.

Data Flow - 4 Stages

1Raw Text Input

1 text string→Input raw sentence for entity recognition→1 text string

"Apple is looking at buying U.K. startup for $1 billion"

↓

2Tokenization

1 text string→Split text into tokens (words and punctuation)→1 list of tokens

["Apple", "is", "looking", "at", "buying", "U.K.", "startup", "for", "$", "1", "billion"]

↓

3NER Model Processing

1 list of tokens→Apply spaCy's NER model to label tokens with entity types→1 list of tokens with entity labels

[("Apple", "ORG"), ("U.K.", "GPE"), ("$", "MONEY"), ("1", "MONEY"), ("billion", "MONEY")]

↓

4Output Entities

1 list of tokens with entity labels→Extract and output recognized entities with their types→List of entities with labels

[{"text": "Apple", "label": "ORG"}, {"text": "U.K.", "label": "GPE"}, {"text": "$1 billion", "label": "MONEY"}]

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	Model starts learning to recognize entities, loss is high, accuracy low.
2	0.60	0.75	Loss decreases, accuracy improves as model learns entity patterns.
3	0.45	0.82	Model shows better entity recognition, loss continues to drop.
4	0.35	0.88	Training converges, accuracy approaches high performance.
5	0.30	0.90	Final epoch with good accuracy and low loss.

Prediction Trace - 4 Layers

Layer 1: Input Text

Layer 2: Tokenization

Layer 3: NER Model

Layer 4: Entity Extraction

Model Quiz - 3 Questions

What does the tokenization stage do in the NER pipeline?

A Extracts entities from labeled tokens

B Labels tokens with entity types

C Splits text into smaller pieces like words and punctuation

D Trains the model to recognize entities

Key Insight

Named Entity Recognition models learn to identify meaningful words or phrases like names and places by training on labeled text. Over time, the model improves by reducing errors (loss) and correctly labeling more entities (accuracy). This pipeline shows how raw text is transformed step-by-step into structured entity information.