witfame

---

1. What is Natural Language Processing (NLP)?

✅ Answer:

Natural Language Processing (NLP) is a subfield of Artificial Intelligence (AI) that focuses on enabling computers to understand, interpret, and generate human language in a meaningful way. NLP combines computational linguistics with machine learning to process and analyze large amounts of natural language data.

🌐 Example:

• Spam Detection: Email services use NLP to identify spam emails based on text patterns and keywords.
• Chatbots: Virtual assistants like Siri or Alexa use NLP to understand and respond to user queries.

2. What are the main challenges in NLP?

✅ Answer:

NLP faces several challenges due to the complexity of human language:

1. Ambiguity: Same word/sentence can have multiple meanings.
2. Sarcasm and Irony: Hard to detect emotional tone.
3. Context Understanding: A word’s meaning can depend on the surrounding words.
4. Data Sparsity: Lack of sufficient training data for some languages or dialects.

🌐 Example:

• “I saw the man with the telescope.”
  → Ambiguity: Did the person see the man using the telescope or see the man holding a telescope?

---

3. What is Tokenization in NLP? Why is it important?

✅ Answer:

Tokenization is the process of splitting text into smaller units called tokens (words, phrases, or characters). It is essential for analyzing the structure and meaning of a sentence.

🌐 Example:

Input:
“I love natural language processing!”

Tokenized Output:
["I", "love", "natural", "language", "processing", "!"]

---

4. What is Lemmatization and Stemming in NLP? How are they different?

✅ Answer:

• Stemming: Reduces words to their root form by chopping off suffixes.
• Lemmatization: Converts a word to its dictionary form using context and grammar rules.

🌐 Example:

Word	Stemming	Lemmatization
Running	run	run
Studies	studi	study
Better	better	good

Key Difference: Lemmatization is more accurate but computationally expensive; stemming is faster but less accurate.

---

5. Explain Named Entity Recognition (NER) with an example.

✅ Answer:

NER is an NLP technique used to identify and classify proper names (entities) in text into predefined categories like Person, Location, Organization, Date, etc.

🌐 Example:

Input:
“Elon Musk founded SpaceX in 2002 in California.”

NER Output:

• Person: Elon Musk
• Organization: SpaceX
• Date: 2002
• Location: California

---

6. What is Part-of-Speech (POS) Tagging?

✅ Answer:

POS tagging is the process of assigning parts of speech (noun, verb, adjective, etc.) to each word in a sentence based on its context and definition.

🌐 Example:

Input:
“The cat sat on the mat.”

POS Output:

• The → Determiner
• cat → Noun
• sat → Verb
• on → Preposition
• the → Determiner
• mat → Noun

---

7. What is the Difference Between Bag of Words (BoW) and TF-IDF?

✅ Answer:

• Bag of Words (BoW): Represents text as a collection of unique words without considering their order.
• TF-IDF (Term Frequency-Inverse Document Frequency): Weighs the importance of words based on their frequency in a document relative to other documents.

🌐 Example:

Input:
“I love NLP. NLP is fun.”

• Bag of Words: {"I": 1, "love": 1, "NLP": 2, "is": 1, "fun": 1}
• TF-IDF: Higher weight for “NLP” if it appears more frequently in one document than across other documents.

---

8. What are Word Embeddings in NLP?

✅ Answer:

Word embeddings are vector representations of words where similar words have similar vector representations. They capture semantic relationships between words.

🌐 Example:

“King – Man + Woman = Queen”

• Word2Vec, GloVe, and FastText are popular word embedding models.

---

9. What is Attention Mechanism in NLP?

✅ Answer:

Attention allows models to focus on relevant parts of the input sequence, improving context understanding, especially in tasks like translation and summarization.

🌐 Example:

In a translation task, the model “attends” to different parts of the sentence to understand the structure.

---

10. What is the Transformer Model in NLP?

✅ Answer:

The Transformer model is a neural network architecture introduced by Google in 2017. It uses self-attention and positional encoding to process sequences efficiently.

🌐 Example:

• BERT (Bidirectional Encoder Representations from Transformers)
• GPT (Generative Pre-trained Transformer)

---

11. What is the Difference Between Statistical NLP and Deep Learning-based NLP?

✅ Answer:

1. Statistical NLP:
   • Based on mathematical models and statistical analysis.
   • Techniques: N-grams, Hidden Markov Models (HMM), POS tagging, TF-IDF.
   • Works well with small datasets but struggles with complex language patterns.
2. Deep Learning-based NLP:
   • Uses neural networks to model language structure and context.
   • Techniques: RNN, LSTM, Transformer, BERT, GPT.
   • Requires large datasets and high computational power but achieves higher accuracy.

🌐 Example:

Task	Statistical NLP	Deep Learning
Spam Detection	Based on word frequency (e.g., ‘win’, ‘free’)	Based on context and language patterns
Translation	Rule-based or phrase-based	Context-based (Google Translate)
Sentiment Analysis	Counting positive/negative words	Understanding context and tone

12. What is Sequence-to-Sequence (Seq2Seq) in NLP?

✅ Answer:

Sequence-to-Sequence models are used to convert one sequence (e.g., text) into another sequence (e.g., translated text). It consists of two main components:

1. Encoder: Processes the input sequence and generates a context vector.
2. Decoder: Uses the context vector to generate the output sequence.

🌐 Example:

Machine Translation:

• Input: “How are you?”
• Output: “¿Cómo estás?”

---

13. What is a Language Model (LM)? How does it work?

✅ Answer:

A Language Model (LM) predicts the probability of a sequence of words. It assigns higher probabilities to grammatically and contextually correct sentences.

🌐 Example:

1. Unigram Model: Probability of individual words.
2. Bigram Model: Probability of two words appearing together.
3. Trigram Model: Probability of three words appearing together.

Sentence:
“I love natural language processing.”

• Unigram → P("I") * P("love") * P("natural") * P("language") * P("processing")
• Bigram → P("I love") * P("love natural") * P("natural language") * P("language processing")

---

14. What is the Difference Between RNN and LSTM?

✅ Answer:

1. Recurrent Neural Networks (RNN):
   • Designed for sequential data (text, speech).
   • Suffers from vanishing gradient problem with long sequences.
2. Long Short-Term Memory (LSTM):
   • Overcomes the vanishing gradient problem using forget gates and cell states.
   • Better at learning long-term dependencies.

🌐 Example:

RNN: “The cat sat on the mat.” → Works well for short sentences.
LSTM: “Once upon a time, there was a king who ruled the kingdom…” → Works well for longer sequences.

---

15. What is Word2Vec and How Does it Work?

✅ Answer:

Word2Vec is a neural network-based model that creates vector representations of words. It uses two main approaches:

1. Continuous Bag of Words (CBOW): Predicts a word from surrounding context.
2. Skip-Gram: Predicts surrounding context from a target word.

🌐 Example:

“King – Man + Woman = Queen”
→ Similar words have closer vectors in space.

---

16. Explain the Difference Between Generative and Discriminative Models in NLP.

✅ Answer:

Type	Description	Example
Generative Model	Learns the joint probability (P(X, Y)) and generates new data.	GPT, LDA
Discriminative Model	Learns conditional probability (P(Y	X)) and makes predictions.

🌐 Example:

• GPT: Completes sentences or writes articles (Generative).
• BERT: Predicts the sentiment of a sentence (Discriminative).

---

17. What is the Difference Between Context-Free Grammar (CFG) and Contextual Grammar?

✅ Answer:

1. Context-Free Grammar (CFG):
   • Rules are applied regardless of context.
   • Limited to simple sentence structures.
2. Contextual Grammar:
   • Considers surrounding words and context.
   • Handles complex sentence structures.

🌐 Example:

• CFG: “The cat sat on the mat.” → Simple syntax-based rule.
• Contextual Grammar: “I know that he knows.” → Meaning depends on context.

---

18. What is Beam Search in NLP? Why is it Important?

✅ Answer:

Beam Search is a decoding algorithm that selects the top-k most likely next words at each step instead of just the highest probability. It balances between exploration and exploitation.

🌐 Example:

Sentence Completion:

• Input: “I am feeling…”
• Beam Search:
  • “I am feeling happy.”
  • “I am feeling tired.”
  • “I am feeling great.”

---

19. What is Text Summarization? What are the Types?

✅ Answer:

Text Summarization generates a shorter version of a text while preserving key information.

1. Extractive: Selects key sentences from the text.
2. Abstractive: Generates a summary in its own words.

🌐 Example:

Input:
“Natural language processing enables computers to understand human language.”

Extractive:
“Computers understand human language.”
Abstractive:
“NLP helps computers comprehend human language.”

20. What is the BLEU Score in NLP? Why is it Important?

✅ Answer:

BLEU (Bilingual Evaluation Understudy) is a metric for evaluating the quality of machine-generated text by comparing it with a reference text.

🌐 Example:

Reference:
“The cat sat on the mat.”
Prediction:
“The cat is sitting on the mat.”
→ BLEU measures how similar the prediction is to the reference text.

✅ More Real-Life Examples and Use Cases:

Use Case	Description
Sentiment Analysis	Analyzing customer reviews to detect positive or negative sentiment.
Language Translation	Translating documents from English to Spanish.
Speech Recognition	Converting spoken words into text (e.g., Siri).
Text Classification	Spam detection in email.
Question Answering	Chatbots like ChatGPT.
Document Clustering	Grouping similar articles together.

✅ Final Pro Tips:

✔️ Be prepared to code simple NLP tasks using libraries like NLTK, Spacy, and HuggingFace.

✔️ Provide clear, structured answers.

✔️ Use practical examples based on real-life applications.

✔️ trade-offs between different models and methods.

✔️ Prepare coding-based questions using NLTK, Spacy, and HuggingFace.
✔️ Be prepared to handle data preprocessing questions.
✔️ Know the trade-offs between different models and techniques.