Brief history of NLP:

1. What is Natural Language?

o Natural languages are languages spoken by people (like English, Hindi, Marathi).
o These are different from artificial/programming languages (like C, C++, Java).
2. What is NLP?
o NLP stands for Natural Language Processing.
o It’s a field of research focused on making computers understand and process
human language, whether text or speech, to perform useful tasks.
3. NLP and Artificial Intelligence (AI):
o NLP is a subset of Artificial Intelligence (AI).
o Its goal is to bridge the communication gap between humans and computers.
o It started with Machine Translation (MT) in the 1950s to translate one human
language into another (e.g., Russian to English).
4. Evolution of NLP:
o In the 1950s: Machine Translation (MT) began, initially focusing on translating
between English and Russian.
o Later, there was a focus on translating human language to computer language and
vice versa.
5. What Does NLP Do?
o NLP takes human language input (e.g., speech or text) and converts it into a
format computer can work with.
o It helps computers perform tasks related to language, like understanding and
responding to questions, translating languages, etc.
6. Key Historical Milestones:
o 1960s: NLP advanced with the rise of Artificial Intelligence (AI).
o 1978: W.A. Woods developed LUNAR, a system that analyzed data from lunar
rock and soil from the Apollo missions and answered questions about it.
o 1980s: Computational grammar became an active research area. Researchers
explored how to make machines understand language meaning, user beliefs, and
intentions.
o 1990s: NLP grew rapidly with the development of tools, grammars, and resources
for analyzing language. Probabilistic models (data-driven approaches) became
standard.
7. Modern NLP (2000s and beyond):
o With the availability of vast spoken and written data, NLP made great advances.
o Today, NLP uses Machine Learning (ML) and Deep Neural Networks (DNN) to
build powerful systems.
o Applications include text classification, question-answer generation, sentiment
analysis, and much more.

NLP has grown from simple machine translation to complex AI-powered systems capable of
understanding and processing human language in diverse ways.
Application of NLP:
Machine Translation:

Converts text or speech from one language to another.

Ensures that the meaning is preserved in the translation.

Example: Google Translator.

Information Retrieval (IR):

Refers to human-computer interaction where a machine searches through a body of information

(like documents or web pages) based on a user’s query.

Finds relevant documents that match the search query.

Examples: Google, Yahoo, Altavista.

Text Categorization (Text Classification):

Automatically sorts a set of documents into categories (also known as clustering).

Uses:

Content filtering

Spam detection

Survey coding Document content identification

Information Extraction (IE):

Identifies specific pieces of information from unstructured or semi-structured text and organizes
them into a structured format (like a database).

Applied to different types of text like newspaper articles, web pages, scientific papers, classified
ads, and medical notes.

Grammar Correction:

Used in word processors like MS Word for spelling correction and grammar checks.

Sentiment Analysis (Opinion Mining):

- Analyzes a person’s attitude, emotions, and feelings based on text.

- Identifies whether the text expresses a positive, negative, or neutral sentiment.

- Commonly used to analyze social media posts, reviews, and online feedback.

Question-Answering Systems (QA):

Systems that directly answer questions asked in natural language.

Provides precise answers instead of full documents like search engines.

Example: Q. “What is the birthplace of Lord Krishna?” A. “Mathura.”

Spam Detection:

Detects and filters unwanted or spam emails from a user's inbox.

Chatbots:

Used by companies to provide automated customer service through conversations.

A key application of NLP in customer support.

Speech Recognition:

Converts spoken words into text.

Used in applications like voice assistants (e.g., Siri, Google Assistant), home automation, and
voice biometrics.

Text Summarization:

Automatically creates brief summaries of long documents while maintaining the key ideas and
meaning
Challenges of NLP:
Language Differences:
o Different languages have unique vocabulary, phrasing, and cultural
nuances.
o “Universal” models can help transfer knowledge across languages, but you
still need to update the NLP system for each new language.
Training Data:
o For NLP to be effective, the system needs lots of high-quality training data
(texts or speech).
o Poor data leads to poor results, so feeding the system with accurate,
relevant data is crucial for success.
Development Time:
o Building an NLP system takes time, especially since AI must process
many data points.
o Using shared deep networks and GPUs can speed up the process, but
developing a strong NLP system still requires significant time and
resources.
Phrasing Ambiguities:
o Sometimes, the meaning of a phrase is unclear or ambiguous, even for
humans.
o NLP systems must understand context or ask for clarification to interpret
such phrases correctly.
Misspellings:
o Humans often make spelling mistakes, but while another human can
usually understand them, machines might struggle.
o NLP systems need tools to detect and correct spelling errors.
Words with Multiple Meanings:
o Many words have different meanings depending on context. For example,
"bank" can refer to a financial institution or the side of a river.
o NLP systems must use context to determine the correct meaning.
2. Phrases with Multiple Intentions:
o Sometimes, a phrase can have multiple requests or actions within it.
o The system should be able to recognize and process each intention
separately (e.g., “cancel my order and update my card”).
3. False Positives and Uncertainty:
o A false positive happens when the system thinks it understands a word but
provides an incorrect answer.
o Good NLP systems need to recognize when they don’t understand
something and ask for more information.
Keeping a Conversation Moving:
o When communicating with humans, NLP systems should keep the
conversation flowing smoothly.
o They should ask clarifying questions or collect more information to guide
the interaction toward a solution.
 These challenges highlight the complexity of building NLP systems that can understand
and respond to human language effectively.
The process of Natural Language Processing is divided into 5 major stages or phases, starting
from basic word-level processing up to finding complex meanings of sentences.
Stages of Natural Language Processing:

1. Morphological Analysis/ Lexical Analysis

Morphological or lexical analysis focuses on individual words, identifying morphemes (e.g.,

prefix, root, suffix) and converting words to their root forms. It also determines possible parts
of speech (POS) for a word, using a language's dictionary for context. For example,
irrationally can be split into ir (prefix), rational (root), and -ly (suffix), while the word
character could serve as either a noun or a verb.

2. Syntax Analysis
Syntax analysis validates the structural correctness of a sentence by parsing it to ensure proper
grammar. Based on the POS tags generated in the previous step, it assigns tags according to the
sentence's structure.

For example:

 Correct Syntax: Sun rises in the east.

 Incorrect Syntax: Rise in sun the east.

3. Semantic Analysis
Semantic analysis focuses on the meaning of a sentence, even if it is syntactically correct. It
ensures logical consistency and combines words into meaningful phrases.

For example:

 The sentence "The apple ate a banana" is syntactically correct but semantically invalid
because apples cannot eat.
 Phrases like "red apple" are treated as a single entity, while names like "Robert Hill" are
understood as referring to one person, not two separate entities.

4. Discourse
Discourse analysis looks at how sentences relate to each other in a larger context, focusing on
how the meaning of one sentence can affect or be connected to the next.

For example:
 1. Sentence 1: "John went to the store." Sentence 2: "He bought some milk."
o Discourse analysis understands that "He" refers to "John."
2. Sentence 1: "Sarah loves to paint." Sentence 2: "Her paintings are displayed in the
gallery."
o Discourse analysis links "Her" to "Sarah" and "paintings" to the act of painting.

Discourse helps maintain coherence by establishing references and relationships across

sentences.

5. Pragmatics
The final stage of NLP, Pragmatics interprets the given text using information from the previous
steps.

Given a sentence, “Turn off the lights” is an order or request to switch off the lights.