Categorisation

Categorisation is the process of organizing items into groups, or categories, based on shared properties or relationships. It helps people and systems to interpret, store, and retrieve information, and to reason about similarities and differences. While categorisation and classification are often used interchangeably, some distinctions exist: categorisation focuses on forming groups, whereas classification emphasizes placing items into preexisting categories.

Categories can be hierarchical (a tree of categories) or flat; they can be mutually exclusive or overlapping; and they may be exhaustive or non-exhaustive depending on context. Criteria for grouping may be theoretical definitions, observable features, or functional roles. In information science and library work, formal systems such as taxonomies and ontologies provide structured categorisation, while folksonomies or tags offer informal, user-generated categorisations. In data science, categorisation includes supervised learning (labeling items with known categories) and unsupervised learning (discovering groupings via clustering).

Human categorisation reflects cognitive processes and varies across individuals and cultures. Theories range from feature-based approaches (using diagnostic attributes) to prototype or exemplar models (comparing to typical or remembered instances).

Applications include library catalogs, product categorisation in commerce, image and document tagging, and content recommendation. In biology, categorisation underpins taxonomy and phylogeny; in linguistics, semantic categorisation groups words by meaning.

Challenges include ambiguous or context-dependent boundaries, polysemy, cultural variation in what counts as a category, and the dynamic evolution of categories over time. When categorisation is automated, evaluation relies on metrics such as accuracy, precision, recall, and F1, and on human judgment for interpretability and fairness.