Please enable JavaScript.
Coggle requires JavaScript to display documents.
Natural Language Processing, Study of human speech sounds., Morphology,…
Natural Language Processing
NLP is the intersection of computer science, linguistics, and psychology that allows computers to interpret, generate, and respond to human language naturally.
Introduction to NLP
Applications
Machine Translation, Speech Recognition, Virtual Assistants.
History of NLP
1950s
: Turing Test (AI intelligence test), Georgetown Experiment (early machine translation).
1960s
: ELIZA chatbot (psychological therapy simulation).
1970s
: Emergence of Conceptual Ontologies, chatterbots like ALICE and Jabberwacky.
1980s
onwards: Machine learning replaces rule-based NLP; development of probabilistic models.
Components of NLP
Linguistics
Syntax
the study of how words are arranged to form correct and meaningful sentences
Helps NLP systems understand sentence structure and grammar
Semantics
Meaning of words/sentences
Pragmatics
Meaning beyond the literal context
how speech sounds are organized and used in different languages
Phonology
Computer Science
Data Structures and Algorithms
efficiently store and process language data, such as words, sentences, and grammar rules
Computational Models
mathematical models that help computers make probabilistic predictions about language, such as the likely next word in a sentence, by learning patterns from large text data
Eg
:
Hidden Markov Models
Cognitive Psychology
Language reflects cognitive processes and thought patterns.
NLP uses psychological theories to model realistic language use.
Language Types
Formal Languages
Strict grammar (programming languages like Python, Java)
Natural Languages
Eg
: English, French
Example
"
He saw her duck
" As apparent, it has multiple interpretations
Practical Examples and Resources
Knowledge Bases
Open Mind Common Sense
A project that gathers everyday human knowledge to help machines understand common sense
ConceptNet
A knowledge graph that connects related concepts to support reasoning in NLP tasks.
Competitions
Loebner Prize
A yearly Turing Test challenge where AI chatbots compete to appear most human in conversation.
Chatbots
Early AI conversational agents designed to simulate human dialogue
Eg
: ELIZA - mimicked a psychotherapist,
ALICE - used pattern-matching
Jabberwacky - aimed for more natural, humorous conversation.
NLP Challenges
Words/sentences having multiple meanings
Understanding common sense
Decoding meaning beyond text
Techniques in NLP
Rule-based Systems
Use manually written rules (if-then logic) to process language
Eg
:
A rule like “If a sentence ends in a question mark, classify it as a question.”
Deep Learning
A subset of ML that uses neural networks with many layers to model complex patterns.
Can automatically learn features
Common models:
Transformers, RNNs, and CNNs
Solved vs Ongoing Problems
Mostly Solved
Optical Character Recognition. Spam Detection (ML), Named Entity Recognition
Improving
Sentiment Analysis, Machine Transition, Parsing
Difficult
Dialogue systems, Natural Language Understanding, Question Answering
NLP Applications
Machine Translation
Google Translate, DeepL
Speech Recognition
Alexa, Siri
Sentiment Analysis
Social media monitoring
Eg
: trends on Twitter
Information Extraction
Structured databases from raw text
Coreference Resolution
Resolving "he," "she" pronouncs to named individuals
Word Sense
Understanding word context meanings
Future Directions
Better context-awareness in NLP
Expansion of NLP tools into sectors like healthcare and education.
Eg
:
extract info from medical records, grading
Integration of comprehensive common-sense knowledge
Advanced dialogue agents
Eg
:
Human identical speech from chatbots
Study of human speech sounds.
Phonetics
Morphology
the study of the structure of words and how they are formed using morphemes
Morphemes are the smallest units of meaning in a language
helps computers break down and understand new or complex words
Machine Learning
Statistical/probabilistic models trained on large datasets.
Named Entity Recognition
Recognizing locations, names in text
Eg
:
Predicting the next word in a sentence based on frequency patterns in training data