Graph View

Natural Language Processing

Natural Language Processing (NLP)

Natural Language Processing (NLP) is a major subfield of artificial-intelligence (AI), computer science, and computational linguistics. It focuses on the interactions between computers and human languages, specifically how to program computers to process, analyze, and generate large amounts of natural language data.

Key Features & Core Tasks

Natural Language Processing involves a wide variety of tasks that enable machines to understand and generate text. These are broadly categorized into Natural Language Understanding (NLU) and Natural Language Generation (NLG):

  • Text Classification & Sentiment Analysis: Categorizing text into predefined classes (e.g., spam detection) or determining the emotional tone (e.g., positive, negative, neutral) of a passage.
  • Named Entity Recognition (NER): Identifying and classifying key entities within text, such as names of people, organizations, locations, dates, or quantities.
  • Part-of-Speech (POS) Tagging: Identifying the grammatical role of each word (e.g., noun, verb, adjective) in a sentence.
  • Machine Translation: Automatically converting text or speech from one natural language to another.
  • Text Summarization: Condensing long documents into shorter versions while preserving key semantic information.
  • Speech Recognition: Converting spoken language into written text, bridging speech processing and textual NLP.
  • Question Answering: Automatically answering questions posed in natural language by querying unstructured text or structured databases.

Key Techniques

NLP systems leverage several core techniques to preprocess text and extract features:

  1. Preprocessing: Includes tokenization (splitting text into words/sub-words), stop word removal (filtering out common filler words like “the” or “is”), and lemmatization/stemming (reducing words to their root dictionary form).
  2. Feature Representation: Converting raw text into numerical representations. Traditional methods include Bag-of-Words and TF-IDF, while modern methods use dense vector representations called Word Embeddings (e.g., Word2Vec, GloVe) and contextual embeddings (e.g., BERT).
  3. Algorithmic Modeling: Transitioning from early rule-based systems (hand-coded grammars) to statistical machine learning models (HMMs, CRFs, SVMs) and ultimately to modern deep-learning neural networks (RNNs, LSTMs, and transformer-models).

Evolution & History

The evolution of NLP is typically divided into three major eras:

  1. Symbolic & Rule-Based Era (1950s–1980s): Characterized by hand-crafted linguistic rules. Key milestones include Alan Turing’s Turing Test proposal (1950), the Georgetown-IBM experiment (1954), and Joseph Weizenbaum’s conversational chatbot ELIZA (1966).
  2. Statistical Era (1990s–2000s): Fueled by increased computing power and digital text availability, the field shifted to probabilistic models that learned patterns from data rather than relying on human-coded rules.
  3. Neural & Transformer Era (2010s–Present): Revolutionized by neural networks and the invention of the Transformer architecture (2017) with its self-attention mechanism. This breakthrough paved the way for large-language-models (LLMs) such as GPT and BERT, which scale to billions of parameters and display emergent conversational and reasoning capabilities.
Properties1
tagsnatural-language-processing, nlp, artificial-intelligence, machine-learning, deep-learning

Backlinks