The three terms — taxonomy, thesaurus, and ontology — are often used interchangeably in knowledge management discussions, but they represent fundamentally different tools. Choosing the right one depends on what you need to do with your knowledge, and starting with the wrong one wastes considerable time and effort.
What a Taxonomy Provides
A taxonomy is a hierarchical classification system. It organises concepts into parent-child relationships, forming a tree structure. Taxonomies answer one question well: what kind of thing is this? They are fast to build, easy to maintain, and immediately useful for faceted browsing and high-level categorisation. Their limitation is that they capture only one dimension of knowledge — the hierarchical one. A drug might belong to one therapeutic class in a taxonomy but relate to dozens of diseases, mechanisms, targets, and adverse effects that a pure taxonomy cannot express.
What a Thesaurus Adds
A thesaurus extends a taxonomy with synonym relationships, related term links, and scope notes. In healthcare, a controlled vocabulary like MeSH is essentially a thesaurus: it tells you not only that neoplasm is narrower than disease, but also that it is equivalent to tumour in many contexts. Thesauri are valuable for information retrieval, helping search systems expand queries to cover synonyms and related concepts. Their limitation is that they do not formally define how concepts relate — "related to" is not a semantically precise relationship.
What an Ontology Enables
An ontology provides formal, typed, machine-interpretable relationships. You can assert that a drug treats a disease, that a disease affects an anatomical region, that a clinical trial investigates a drug. These relationships have formal semantics — they can be combined, inverted, and reasoned over by inference engines. A thesaurus might tell you that metformin and type 2 diabetes are related. An ontology asserts that metformin has-indicated-use type-2-diabetes-mellitus, and a reasoner can answer "find all drugs with indicated use in conditions characterised by insulin resistance" correctly, without a human having pre-entered that answer.
Many mature knowledge programmes employ all three in layers: a taxonomy for navigation, a controlled vocabulary thesaurus for search, and a formal ontology for reasoning. The key insight is to start with the simplest tool that solves your current problem — and to design it so that upgrading to the next level does not require discarding what you have already built.
