Text Classification: Definition and Examples

Text classification is one of the fundamental tasks of natural language processing (NLP). It involves analyzing a text — whether an email, customer review, article, or tweet — and automatically assigning it a label from a set of predefined categories. For example, classifying an email as "spam" or "not spam," or determining whether a comment expresses a positive, negative, or neutral sentiment.

Historically, text classification relied on statistical approaches such as Naive Bayes models or SVMs (support vector machines), which required significant feature engineering. With the advent of deep learning and language models like BERT, then large generative models (LLMs), text classification has undergone a revolution: it is now possible to classify texts with high precision using simply a well-crafted prompt, without training a specific model.

In the context of prompt engineering, text classification is one of the most common and accessible use cases. One can ask an LLM to classify a text by providing the desired categories directly in the prompt, using techniques such as zero-shot (without examples), few-shot (with a few examples), or chain-of-thought (by asking the model to explain its reasoning before deciding).

Practical applications are vast: content moderation, automatic routing of support tickets, sentiment analysis, document categorization, intent detection in chatbots, or automatic email sorting. The quality of classification depends heavily on the clarity of the defined categories and the precision of the instructions given in the prompt.