Natural Language Processing
Natural Language Processing, or NLP, is the subfield of AI concerned with enabling computers to understand, interpret, and generate human language. Languag...
Understanding Human Language
Natural Language Processing, or NLP, is the subfield of AI concerned with enabling computers to understand, interpret, and generate human language. Language is extraordinarily complex. The same sentence can mean different things depending on context, tone, and shared background knowledge. Words are ambiguous, grammar is irregular, and meaning shifts across cultures and time. NLP must address all of these challenges to be useful in practice.
Classic NLP pipelines broke language processing into discrete stages: tokenisation splits text into individual words or subwords, part-of-speech tagging labels each token as a noun, verb, adjective, and so on, named entity recognition identifies people, places, and organisations, parsing builds a structural tree showing grammatical relationships, and coreference resolution links pronouns to the entities they refer to. Each stage was traditionally handled by a separate module with its own trained model.
The modern transformer architecture, introduced in the 2017 paper "Attention Is All You Need," revolutionised NLP by processing all tokens in a sequence in parallel using a mechanism called self-attention, which allows each token to consider its relationship to every other token simultaneously. Large language models built on this architecture, such as BERT for understanding tasks and GPT for generation tasks, achieve state-of-the-art performance across almost every NLP benchmark by pre-training on vast amounts of text and then fine-tuning on specific tasks.
A Practical Example
Sentiment analysis is one of the most common NLP applications. A company can automatically classify millions of product reviews as positive, negative, or neutral rather than paying analysts to read each one. A model trained on labelled reviews learns that phrases like "absolutely love" signal positive sentiment and "complete waste of money" signals negative sentiment.
Exam Tip
NLP questions often test knowledge of ambiguity types. Lexical ambiguity arises when a word has multiple meanings, like "bank" meaning a riverbank or a financial institution. Syntactic ambiguity arises when a sentence can be parsed in multiple ways, the classic example being "I saw the man with the telescope." Being able to give examples of each type and explain how modern models handle them using context is a strong exam answer.