Module 7 · 100 Days of NLP

Module 7: Part-of-Speech Tagging & Hidden Markov Models

Label syntactic parts-of-speech: model lexical tagging via Hidden Markov Models transition/emission likelihood matrices, solved using the Viterbi decoding algorithm.

⏱ 30 Min Read • Author: GenAIWallah Team • Updated: May 2026

Day 66

POS Tagging Intro

Why this matters

POS Tagging Intro: POS, NER, and parsing are sequence labeling — HMMs and CRFs teach the probabilistic view.

POS Tagging Intro is a core topic in the 100 Days of NLP curriculum. This lesson connects theory to practical pipelines you will build in projects.

Sequence labeling

Key takeaways

Define POS Tagging Intro clearly and state when to use it.
Connect this topic to the previous and next day in the curriculum.
Validate with a small code experiment or worked numeric example.

Common mistakes

Confusing transition vs emission probabilities in HMMs.
Applying generative HMM scores as if they were calibrated probabilities.
Ignoring OOV words in Viterbi decoding.

Interview checkpoints

Q: Explain pos tagging intro in one minute. A: State definition, when to use it, and one failure mode.
Q: How does pos tagging intro fit in an NLP pipeline? A: Name inputs, outputs, and what breaks if this step is wrong.

Practice

Basic: Define POS Tagging Intro and give one real product example.
Intermediate: Implement or sketch a minimal example for POS Tagging Intro.
Advanced: Compare POS Tagging Intro to the previous topic on the same dataset.

Recap

You can explain pos tagging intro clearly.
You know one common mistake and how to avoid it.
You see how this connects to the next topic.

Next: Penn Treebank Tags

Day 67

Penn Treebank Tags

Why this matters

Penn Treebank Tags: This NLP concept connects theory to the models and APIs you will use in projects.

Penn Treebank Tags is a core topic in the 100 Days of NLP curriculum. This lesson connects theory to practical pipelines you will build in projects.

Sequence labeling

Key takeaways

Define Penn Treebank Tags clearly and state when to use it.
Connect this topic to the previous and next day in the curriculum.
Validate with a small code experiment or worked numeric example.

Common mistakes

Skipping train/validation split discipline.
Ignoring inference latency and memory.
No error analysis on misclassified examples.

Interview checkpoints

Q: Explain penn treebank tags in one minute. A: State definition, when to use it, and one failure mode.
Q: How does penn treebank tags fit in an NLP pipeline? A: Name inputs, outputs, and what breaks if this step is wrong.

Practice

Basic: Define Penn Treebank Tags and give one real product example.
Intermediate: Implement or sketch a minimal example for Penn Treebank Tags.
Advanced: Compare Penn Treebank Tags to the previous topic on the same dataset.

Recap

You can explain penn treebank tags clearly.
You know one common mistake and how to avoid it.
You see how this connects to the next topic.

Next: HMM Theory

Day 68

HMM Theory

Why this matters

HMM Theory: POS, NER, and parsing are sequence labeling — HMMs and CRFs teach the probabilistic view.

HMM Theory is a core topic in the 100 Days of NLP curriculum. This lesson connects theory to practical pipelines you will build in projects.

Sequence labeling

HMM joint probability

$$P(\mathbf{y}, \mathbf{x}) = P(y_1) \prod_{t=2}^{T} P(y_t \mid y_{t-1}) \prod_{t=1}^{T} P(x_t \mid y_t)$$

Key takeaways

Define HMM Theory clearly and state when to use it.
Connect this topic to the previous and next day in the curriculum.
Validate with a small code experiment or worked numeric example.

Common mistakes

Confusing transition vs emission probabilities in HMMs.
Applying generative HMM scores as if they were calibrated probabilities.
Ignoring OOV words in Viterbi decoding.

Interview checkpoints

Q: Explain hmm theory in one minute. A: State definition, when to use it, and one failure mode.
Q: How does hmm theory fit in an NLP pipeline? A: Name inputs, outputs, and what breaks if this step is wrong.

Practice

Basic: Define HMM Theory and give one real product example.
Intermediate: Implement or sketch a minimal example for HMM Theory.
Advanced: Compare HMM Theory to the previous topic on the same dataset.

Recap

You can explain hmm theory clearly.
You know one common mistake and how to avoid it.
You see how this connects to the next topic.

Next: HMM Transitions

Day 69

HMM Transitions

Why this matters

HMM Transitions: POS, NER, and parsing are sequence labeling — HMMs and CRFs teach the probabilistic view.

HMM Transitions is a core topic in the 100 Days of NLP curriculum. This lesson connects theory to practical pipelines you will build in projects.

Sequence labeling

Key takeaways

Define HMM Transitions clearly and state when to use it.
Connect this topic to the previous and next day in the curriculum.
Validate with a small code experiment or worked numeric example.

Common mistakes

Confusing transition vs emission probabilities in HMMs.
Applying generative HMM scores as if they were calibrated probabilities.
Ignoring OOV words in Viterbi decoding.

Interview checkpoints

Q: Explain hmm transitions in one minute. A: State definition, when to use it, and one failure mode.
Q: How does hmm transitions fit in an NLP pipeline? A: Name inputs, outputs, and what breaks if this step is wrong.

Practice

Basic: Define HMM Transitions and give one real product example.
Intermediate: Implement or sketch a minimal example for HMM Transitions.
Advanced: Compare HMM Transitions to the previous topic on the same dataset.

Recap

You can explain hmm transitions clearly.
You know one common mistake and how to avoid it.
You see how this connects to the next topic.

Next: HMM Emissions

Day 70

HMM Emissions

Why this matters

HMM Emissions: POS, NER, and parsing are sequence labeling — HMMs and CRFs teach the probabilistic view.

HMM Emissions is a core topic in the 100 Days of NLP curriculum. This lesson connects theory to practical pipelines you will build in projects.

Sequence labeling

Key takeaways

Define HMM Emissions clearly and state when to use it.
Connect this topic to the previous and next day in the curriculum.
Validate with a small code experiment or worked numeric example.

Common mistakes

Confusing transition vs emission probabilities in HMMs.
Applying generative HMM scores as if they were calibrated probabilities.
Ignoring OOV words in Viterbi decoding.

Interview checkpoints

Q: Explain hmm emissions in one minute. A: State definition, when to use it, and one failure mode.
Q: How does hmm emissions fit in an NLP pipeline? A: Name inputs, outputs, and what breaks if this step is wrong.

Practice

Basic: Define HMM Emissions and give one real product example.
Intermediate: Implement or sketch a minimal example for HMM Emissions.
Advanced: Compare HMM Emissions to the previous topic on the same dataset.

Recap

You can explain hmm emissions clearly.
You know one common mistake and how to avoid it.
You see how this connects to the next topic.

Next: Forward Algorithm

Day 71

Forward Algorithm

Why this matters

Forward Algorithm: This NLP concept connects theory to the models and APIs you will use in projects.

Forward Algorithm is a core topic in the 100 Days of NLP curriculum. This lesson connects theory to practical pipelines you will build in projects.

Sequence labeling

Key takeaways

Define Forward Algorithm clearly and state when to use it.
Connect this topic to the previous and next day in the curriculum.
Validate with a small code experiment or worked numeric example.

Common mistakes

Skipping train/validation split discipline.
Ignoring inference latency and memory.
No error analysis on misclassified examples.

Interview checkpoints

Q: Explain forward algorithm in one minute. A: State definition, when to use it, and one failure mode.
Q: How does forward algorithm fit in an NLP pipeline? A: Name inputs, outputs, and what breaks if this step is wrong.

Practice

Basic: Define Forward Algorithm and give one real product example.
Intermediate: Implement or sketch a minimal example for Forward Algorithm.
Advanced: Compare Forward Algorithm to the previous topic on the same dataset.

Recap

You can explain forward algorithm clearly.
You know one common mistake and how to avoid it.
You see how this connects to the next topic.

Next: Viterbi Decoding

Day 72

Viterbi Decoding

Why this matters

Viterbi Decoding: POS, NER, and parsing are sequence labeling — HMMs and CRFs teach the probabilistic view.

Viterbi Decoding is a core topic in the 100 Days of NLP curriculum. This lesson connects theory to practical pipelines you will build in projects.

Sequence labeling

The Viterbi algorithm finds the most likely tag sequence in $O(T \cdot |S|^2)$ using dynamic programming.

Key takeaways

Define Viterbi Decoding clearly and state when to use it.
Connect this topic to the previous and next day in the curriculum.
Validate with a small code experiment or worked numeric example.

Common mistakes

Confusing transition vs emission probabilities in HMMs.
Applying generative HMM scores as if they were calibrated probabilities.
Ignoring OOV words in Viterbi decoding.

Interview checkpoints

Q: Explain viterbi decoding in one minute. A: State definition, when to use it, and one failure mode.
Q: How does viterbi decoding fit in an NLP pipeline? A: Name inputs, outputs, and what breaks if this step is wrong.

Practice

Basic: Define Viterbi Decoding and give one real product example.
Intermediate: Implement or sketch a minimal example for Viterbi Decoding.
Advanced: Compare Viterbi Decoding to the previous topic on the same dataset.

Recap

You can explain viterbi decoding clearly.
You know one common mistake and how to avoid it.
You see how this connects to the next topic.

Next: Baum-Welch Training

Day 73

Baum-Welch Training

Why this matters

Baum-Welch Training: POS, NER, and parsing are sequence labeling — HMMs and CRFs teach the probabilistic view.

Baum-Welch Training is a core topic in the 100 Days of NLP curriculum. This lesson connects theory to practical pipelines you will build in projects.

Sequence labeling

Key takeaways

Define Baum-Welch Training clearly and state when to use it.
Connect this topic to the previous and next day in the curriculum.
Validate with a small code experiment or worked numeric example.

Common mistakes

Confusing transition vs emission probabilities in HMMs.
Applying generative HMM scores as if they were calibrated probabilities.
Ignoring OOV words in Viterbi decoding.

Interview checkpoints

Q: Explain baum-welch training in one minute. A: State definition, when to use it, and one failure mode.
Q: How does baum-welch training fit in an NLP pipeline? A: Name inputs, outputs, and what breaks if this step is wrong.

Practice

Basic: Define Baum-Welch Training and give one real product example.
Intermediate: Implement or sketch a minimal example for Baum-Welch Training.
Advanced: Compare Baum-Welch Training to the previous topic on the same dataset.

Recap

You can explain baum-welch training clearly.
You know one common mistake and how to avoid it.
You see how this connects to the next topic.

Next: NER with HMMs

Day 74

NER with HMMs

Why this matters

NER with HMMs: POS, NER, and parsing are sequence labeling — HMMs and CRFs teach the probabilistic view.

NER with HMMs is a core topic in the 100 Days of NLP curriculum. This lesson connects theory to practical pipelines you will build in projects.

Sequence labeling

Key takeaways

Define NER with HMMs clearly and state when to use it.
Connect this topic to the previous and next day in the curriculum.
Validate with a small code experiment or worked numeric example.

Common mistakes

Confusing transition vs emission probabilities in HMMs.
Applying generative HMM scores as if they were calibrated probabilities.
Ignoring OOV words in Viterbi decoding.

Interview checkpoints

Q: Explain ner with hmms in one minute. A: State definition, when to use it, and one failure mode.
Q: How does ner with hmms fit in an NLP pipeline? A: Name inputs, outputs, and what breaks if this step is wrong.

Practice

Basic: Define NER with HMMs and give one real product example.
Intermediate: Implement or sketch a minimal example for NER with HMMs.
Advanced: Compare NER with HMMs to the previous topic on the same dataset.

Recap

You can explain ner with hmms clearly.
You know one common mistake and how to avoid it.
You see how this connects to the next topic.

Next: CRF for Sequence Labeling

Day 75

CRF for Sequence Labeling

Why this matters

CRF for Sequence Labeling: POS, NER, and parsing are sequence labeling — HMMs and CRFs teach the probabilistic view.

CRF for Sequence Labeling is a core topic in the 100 Days of NLP curriculum. This lesson connects theory to practical pipelines you will build in projects.

Sequence labeling

Key takeaways

Define CRF for Sequence Labeling clearly and state when to use it.
Connect this topic to the previous and next day in the curriculum.
Validate with a small code experiment or worked numeric example.

Common mistakes

Confusing transition vs emission probabilities in HMMs.
Applying generative HMM scores as if they were calibrated probabilities.
Ignoring OOV words in Viterbi decoding.

Interview checkpoints

Q: Explain crf for sequence labeling in one minute. A: State definition, when to use it, and one failure mode.
Q: How does crf for sequence labeling fit in an NLP pipeline? A: Name inputs, outputs, and what breaks if this step is wrong.

Practice

Basic: Define CRF for Sequence Labeling and give one real product example.
Intermediate: Implement or sketch a minimal example for CRF for Sequence Labeling.
Advanced: Compare CRF for Sequence Labeling to the previous topic on the same dataset.

Recap

You can explain crf for sequence labeling clearly.
You know one common mistake and how to avoid it.
You see how this connects to the next topic.

Next: SpaCy POS Tagger

Day 76

SpaCy POS Tagger

Why this matters

SpaCy POS Tagger: POS, NER, and parsing are sequence labeling — HMMs and CRFs teach the probabilistic view.

SpaCy POS Tagger is a core topic in the 100 Days of NLP curriculum. This lesson connects theory to practical pipelines you will build in projects.

Sequence labeling

SpaCy's statistical tagger combines embeddings with transition-based parsing — use doc[i].pos_ for quick experiments.

Key takeaways

Define SpaCy POS Tagger clearly and state when to use it.
Connect this topic to the previous and next day in the curriculum.
Validate with a small code experiment or worked numeric example.

Common mistakes

Confusing transition vs emission probabilities in HMMs.
Applying generative HMM scores as if they were calibrated probabilities.
Ignoring OOV words in Viterbi decoding.

Interview checkpoints

Q: Explain spacy pos tagger in one minute. A: State definition, when to use it, and one failure mode.
Q: How does spacy pos tagger fit in an NLP pipeline? A: Name inputs, outputs, and what breaks if this step is wrong.

Practice

Basic: Define SpaCy POS Tagger and give one real product example.
Intermediate: Implement or sketch a minimal example for SpaCy POS Tagger.
Advanced: Compare SpaCy POS Tagger to the previous topic on the same dataset.

Recap

You can explain spacy pos tagger clearly.
You know one common mistake and how to avoid it.
You see how this connects to the next topic.

Next: Chunking & Parsing

Day 77

Chunking & Parsing

Why this matters

Chunking & Parsing: POS, NER, and parsing are sequence labeling — HMMs and CRFs teach the probabilistic view.

Chunking & Parsing is a core topic in the 100 Days of NLP curriculum. This lesson connects theory to practical pipelines you will build in projects.

Sequence labeling

Key takeaways

Define Chunking & Parsing clearly and state when to use it.
Connect this topic to the previous and next day in the curriculum.
Validate with a small code experiment or worked numeric example.

Common mistakes

Confusing transition vs emission probabilities in HMMs.
Applying generative HMM scores as if they were calibrated probabilities.
Ignoring OOV words in Viterbi decoding.

Interview checkpoints

Q: Explain chunking & parsing in one minute. A: State definition, when to use it, and one failure mode.
Q: How does chunking & parsing fit in an NLP pipeline? A: Name inputs, outputs, and what breaks if this step is wrong.

Practice

Basic: Define Chunking & Parsing and give one real product example.
Intermediate: Implement or sketch a minimal example for Chunking & Parsing.
Advanced: Compare Chunking & Parsing to the previous topic on the same dataset.

Recap

You can explain chunking & parsing clearly.
You know one common mistake and how to avoid it.
You see how this connects to the next topic.

Next: Dependency Parsing

Day 78

Dependency Parsing

Why this matters

Dependency Parsing: POS, NER, and parsing are sequence labeling — HMMs and CRFs teach the probabilistic view.

Dependency Parsing is a core topic in the 100 Days of NLP curriculum. This lesson connects theory to practical pipelines you will build in projects.

Sequence labeling

Key takeaways

Define Dependency Parsing clearly and state when to use it.
Connect this topic to the previous and next day in the curriculum.
Validate with a small code experiment or worked numeric example.

Common mistakes

Confusing transition vs emission probabilities in HMMs.
Applying generative HMM scores as if they were calibrated probabilities.
Ignoring OOV words in Viterbi decoding.

Interview checkpoints

Q: Explain dependency parsing in one minute. A: State definition, when to use it, and one failure mode.
Q: How does dependency parsing fit in an NLP pipeline? A: Name inputs, outputs, and what breaks if this step is wrong.

Practice

Basic: Define Dependency Parsing and give one real product example.
Intermediate: Implement or sketch a minimal example for Dependency Parsing.
Advanced: Compare Dependency Parsing to the previous topic on the same dataset.

Recap

You can explain dependency parsing clearly.
You know one common mistake and how to avoid it.
You see how this connects to the next topic.

Next: Constituency Parsing

Day 79

Constituency Parsing

Why this matters

Constituency Parsing: POS, NER, and parsing are sequence labeling — HMMs and CRFs teach the probabilistic view.

Constituency Parsing is a core topic in the 100 Days of NLP curriculum. This lesson connects theory to practical pipelines you will build in projects.

Sequence labeling

Key takeaways

Define Constituency Parsing clearly and state when to use it.
Connect this topic to the previous and next day in the curriculum.
Validate with a small code experiment or worked numeric example.

Common mistakes

Confusing transition vs emission probabilities in HMMs.
Applying generative HMM scores as if they were calibrated probabilities.
Ignoring OOV words in Viterbi decoding.

Interview checkpoints

Q: Explain constituency parsing in one minute. A: State definition, when to use it, and one failure mode.
Q: How does constituency parsing fit in an NLP pipeline? A: Name inputs, outputs, and what breaks if this step is wrong.

Practice

Basic: Define Constituency Parsing and give one real product example.
Intermediate: Implement or sketch a minimal example for Constituency Parsing.
Advanced: Compare Constituency Parsing to the previous topic on the same dataset.

Recap

You can explain constituency parsing clearly.
You know one common mistake and how to avoid it.
You see how this connects to the next topic.

Next: Sequence Labeling Project

Day 80

Sequence Labeling Project

Why this matters

Sequence Labeling Project: This NLP concept connects theory to the models and APIs you will use in projects.

Sequence Labeling Project is a core topic in the 100 Days of NLP curriculum. This lesson connects theory to practical pipelines you will build in projects.

Sequence labeling

Key takeaways

Define Sequence Labeling Project clearly and state when to use it.
Connect this topic to the previous and next day in the curriculum.
Validate with a small code experiment or worked numeric example.

Common mistakes

Skipping train/validation split discipline.
Ignoring inference latency and memory.
No error analysis on misclassified examples.

Interview checkpoints

Q: Explain sequence labeling project in one minute. A: State definition, when to use it, and one failure mode.
Q: How does sequence labeling project fit in an NLP pipeline? A: Name inputs, outputs, and what breaks if this step is wrong.

Practice

Basic: Define Sequence Labeling Project and give one real product example.
Intermediate: Implement or sketch a minimal example for Sequence Labeling Project.
Advanced: Compare Sequence Labeling Project to the previous topic on the same dataset.

Recap

You can explain sequence labeling project clearly.
You know one common mistake and how to avoid it.
You see how this connects to the next topic.

Next: Next module

← Module 6: Classification Module 8: Quora Case Study →