Phases of Analysis in Natural Language Processing (NLP) with

Examples:
NLP involves several phases of analysis to understand and process human
language. Here’s a simple breakdown with real-time examples:

1. Lexical Analysis (Tokenization)

 What it does: Breaks text into words, phrases, or symbols (tokens).
 Example:
o Input: "I love ice cream!"
o Output: ["I", "love", "ice", "cream", "!"]

2. Syntactic Analysis (Parsing)

 What it does: Checks grammar and sentence structure (subject-verb-object).
 Example:
o Input: "She eats an apple."
o Output:
 Subject: She
 Verb: eats
 Object: an apple

3. Semantic Analysis (Meaning)

 What it does: Extracts the meaning of words and sentences.
 Example:
o Input: "The bank is closed."
o Meaning: "Bank" refers to a financial institution (not a river bank).

4. Pragmatic Analysis (Context Understanding)

 What it does: Interprets meaning based on real-world context.
 Example:
o Input: "Can you open the window?"
o Interpretation: A request (not just a yes/no question).

5. Discourse Integration (Connecting Sentences)

 What it does: Links sentences to understand the full context.
 Example:
o Input: "John was hungry. He ordered pizza."
o Understanding: "He" refers to John.
 6. Named Entity Recognition (NER)
 What it does: Identifies names of people, places, organizations, etc.
 Example:
o Input: "Elon Musk founded SpaceX in California."
o Output:
 Person: Elon Musk
 Organization: SpaceX
 Location: California

Explain Text Normalization & techniques and tasks involved in

text normalization.
Text Normalization in NLP
Text normalization is the process of converting text into a consistent, standardized form to make it easier
for NLP models to process. It involves cleaning and transforming raw text into a uniform format by handling
variations like case sensitivity, punctuation, abbreviations, and misspellings.

Techniques & Tasks in Text Normalization

1. Lowercasing

 Converts all text to lowercase to ensure uniformity.

 Example:
o Input: "Hello World!"
o Output: "hello world!"

2. Removing Punctuation & Special Characters

 Eliminates unnecessary symbols that don’t contribute to meaning.

 Example:
o Input: "Hey, how are you?"
o Output: "Hey how are you"

3. Stopword Removal

 Removes common words (e.g., "the," "is," "and") that add little meaning.
 Example:
o Input: "The cat sat on the mat."
o Output: "cat sat mat"

4. Stemming

 Reduces words to their root form by cutting off prefixes/suffixes (may not always be correct).
 Example:
o Input: "running, runs, ran"
o Output: "run, run, run"

5. Correcting Misspellings & Typos

 Fixes spelling errors using dictionaries or algorithms.

 Example:
o Input: "I luv NLP!"
o Output: "I love NLP!"

Define different Levels of NLP:

Levels of Natural Language Processing (NLP)

NLP involves multiple levels of analysis, each focusing on different aspects of language understanding. These levels help
computers process human language systematically.

1. Phonetic & Phonological Level

 Focus: Sounds and pronunciation in speech.
 Example:
o Recognizing that "knight" (silent 'k') and "night" sound the same.

2. Morphological Level
 Focus: Structure of words (prefixes, suffixes, root words).
 Example:
o Breaking "unhappiness" → "un" (prefix) + "happy" (root) + "ness" (suffix).

3. Lexical Level (Word-Level Analysis)

 Focus: Meaning of individual words (vocabulary).

 Example:
o "Bank" can mean a financial institution or the side of a river.

4. Syntactic Level (Grammar & Sentence Structure)

 Focus: How words combine to form grammatically correct sentences.

 Example:
o Parsing: "She eats an apple" → Subject (She) + Verb (eats) + Object (an apple).

5. Semantic Level (Meaning of Sentences)

 Focus: Understanding the meaning of words and sentences.

 Example:
o "The chicken is ready to eat" → Could mean:
 The chicken (bird) is hungry.
 The chicken (food) is cooked and can be eaten

Define Linguistic constituents and constituency test.

1.1 What is a Constituent?
A constituent is a word or a group of words that function as a single unit within a sentence’s syntactic
structure. Constituents are the building blocks of sentences and can be identified through various linguistic
tests.
1.2 Types of Constituents
Constituents can be categorized based on their syntactic roles:
 a. Lexical Constituents (Single Words)
 Nouns (e.g., dog), verbs (e.g., run), adjectives (e.g., happy), adverbs (e.g., quickly), etc.
 Example: In "Birds fly," both "birds" (noun) and "fly" (verb) are lexical constituents.

b. Phrasal Constituents (Groups of Words)

 Noun Phrase (NP): "The black cat"
 Verb Phrase (VP): "is sleeping"
 Prepositional Phrase (PP): "on the mat"
 Adjective Phrase (AdjP): "very tall"
 Adverb Phrase (AdvP): "quite slowly"

c. Clausal Constituents (Sentences or Sub-Sentences)

 Independent Clause: "She reads books."
 Dependent Clause: "Because she loves learning."

2. Constituency Tests: Methods to Identify Constituents

Constituency tests are empirical methods to determine whether a group of words
functions as a single unit in a sentence. Below are the most widely used tests with
detailed explanations and examples.
2.1 Substitution (Replacement) Test
 Principle: If a group of words can be replaced by a single word (e.g., a pronoun, do
so, there), it is a constituent.
 Types of Substitutions:
o Pronoun Replacement (for NPs):
 Original: "The students solved the problem."
 Test: "They solved the problem." ("The students" → "They")
 Conclusion: "The students" is an NP constituent.
o Pro-Verb Replacement (for VPs):
 Original: "She will buy a car."
 Test: "She will do so." ("buy a car" → "do so")
 Conclusion: "buy a car" is a VP constituent.

Define dependency parsing with good example:

What is Dependency Parsing?

Dependency Parsing is a linguistic analysis technique that identifies grammatical
relationships between words in a sentence. Unlike constituency parsing (which groups words into
phrases), dependency parsing focuses on direct connections between words, where one word
(the head) governs another (the dependent).
Each relationship is represented as a directed link (an arrow) from the head to the dependent,
labeled with the type of grammatical relation (e.g., subject, object, modifier).
 Key Concepts in Dependency Parsing
1. Head (Governor): The word that controls the relationship.
2. Dependent (Child): The word that depends on the head.
3. Dependency Relation: The grammatical role connecting them (e.g., subject, object, modifier).

Example of Dependency Parsing

Sentence:
"The cat chased the mouse."
Step-by-Step Parsing:
1. "chased" → Root (main verb)
o Governs the entire sentence.
2. "cat" → Subject (nsubj) of "chased"
o "cat" is the doer of the action.
3. "The" → Determiner (det) modifying "cat"
o Specifies which cat.
4. "mouse" → Direct Object (dobj) of "chased"
o The thing being chased.
5. "the" → Determiner (det) modifying "mouse"
o Specifies which mouse.

Define Partial or Shallow Parsing with example:

What is Partial Parsing?

Partial Parsing (also called Shallow Parsing or Chunking) is a lightweight syntactic
analysis technique that identifies phrases (chunks) in a sentence without fully
analyzing their internal structure or grammatical relationships.
Unlike full parsing (which builds a complete syntax tree), shallow parsing only:
✅ Segments text into meaningful chunks (e.g., noun phrases, verb phrases).
✅ Does not resolve deep dependencies (e.g., subject-object relationships).

Key Features of Shallow Parsing

1. Faster than full parsing (useful for real-time NLP tasks).
2. Focuses on surface-level structure (ignores complex grammar rules).
3. Works well for noisy/incomplete text (e.g., social media, search queries)

Example of Shallow Parsing

Sentence:
"The quick brown fox jumps over the lazy dog."
 Step 1: Tokenization
Split into words:
["The", "quick", "brown", "fox", "jumps", "over", "the", "lazy", "dog",
"."]

Step 2: Part-of-Speech (POS) Tagging

Assign word categories:
 The/DT quick/JJ brown/JJ fox/NN jumps/VBZ over/IN the/DT lazy/JJ dog/NN ./.

Applications of Shallow Parsing

1. Information Extraction
o Identifies key phrases like [person names], [locations], [dates].
o Example:
 "Meet [John Doe] at [Paris] on [June 5]."
2. Search Engines
o Improves query understanding:
 Query: "best Italian restaurants" → Chunks: [best Italian] [restaurants].
3. Text Summarization
o Extracts key noun phrases to generate summaries.
4. Sentiment Analysis
o Focuses on phrases like [great service], [terrible experience].

Explain Viterbi algorithm.:

Viterbi Algorithm: Definition and Explanation
The Viterbi algorithm is a dynamic programming algorithm used to find the most
likely sequence of hidden states (the Viterbi path) in a Hidden Markov Model
(HMM), given a sequence of observed events. It is widely used in:
 Speech recognition (e.g., converting sounds to words),
 Natural Language Processing (e.g., part-of-speech tagging),
 Bioinformatics (e.g., DNA sequence analysis),
 Error correction in communication systems (e.g., decoding convolutional codes).

Key Concepts
1. Hidden Markov Model (HMM)
o A statistical model where:
 Observations are visible (e.g., words in a sentence).
 Hidden states are unknown (e.g., part-of-speech tags).
o Defined by:
 Transition probabilities (between hidden states).
 Emission probabilities (probability of observations given states).
2. Goal of Viterbi Algorithm
o Find the optimal sequence of hidden states that best explains the observed data.

Why Use the Viterbi Algorithm?

✅ Efficient: Avoids brute-force search by reusing computations (dynamic programming).
✅ Optimal: Guarantees the single best path (unlike beam search, which approximates).
✅ Widely Used: Essential in NLP (POS tagging, named entity recognition), speech
recognition, and decoding problems.

Define Part -of-speech tagging.

1. Definition and Fundamentals
1.1 What is POS Tagging?
Part-of-Speech (POS) tagging is the process of assigning grammatical categories (noun, verb,
adjective, etc.) to each word in a text based on:
 The word's definition
 Its morphological form (prefixes/suffixes)
 Its syntactic context (relationship with neighboring words)

1.2 Why is POS Tagging Important?

 Foundation for higher-level NLP tasks: Parsing, named entity recognition, machine translation
 Disambiguates word meanings: Resolves whether "book" is a noun or verb
 Improves text understanding: Helps machines interpret grammatical structure

Rule-Based Tagging
 Uses handcrafted linguistic rules
 Example: "If word ends with '-ing', tag as VBG"
 Pros: Transparent, works well for regular patterns
 Cons: Labor-intensive, can't handle ambiguity well

Define database of lexical Relation & give an example.

Definition:
A Database of Lexical Relations is a structured collection of words linked by semantic,
syntactic, or morphological relationships. It serves as a computational lexicon for NLP tasks,
enabling machines to understand word associations like synonyms, antonyms, hypernyms
(generalizations), and meronyms (part-whole relationships).

Key Components:
1. Lexical Entries: Words/phrases with linguistic properties (e.g., "dog" = noun).
2. Relation Types: Predefined categories (e.g., synonymy, antonymy).
3. Metadata: Additional info (e.g., word frequency, part-of-speech).
Common Lexical Relations

Relation Description Example

Synonymy Words with similar meanings happy ≈ joyful

Antonymy Words with opposite meanings hot ≠ cold

Hypernymy General → Specific (IS-A) animal → dog

Hyponymy Specific → General rose → flower

Meronymy Part-Whole relationship wheel → car

Holonymy Whole-Part relationship car → wheel

Conclusion
A Database of Lexical Relations formalizes word connections, bridging human language and
machine understanding. Tools like WordNet power:
✅ Thesauri
✅ Grammar checkers
✅ AI language models
By structuring vocabulary into relational networks, these databases enable nuanced NLP
applications.