Words in natural language

As mentioned in section 2.2, concepts are the basic units of KO. Concepts classify the variety of experience in individual notions, which can be related to other concepts in information networks. However, concepts are not material entities. In order to be shared and transmitted in a culture through its media, either oral or recorded, they need to be expressed in some way.

The most common way to express concepts is by words; other symbols, like gestures, icons or any shared code, can also work. In a sense, then, words in natural languages are a basic form of KO: when we say that cat, we are classing the phenomenon we observe as an instance of the class of cats.

Words are listed in dictionaries with a definition that describes their meaning. Cat means ‘a small carnivoran mammal with tail …’, etc. Such definitions are intended to enumerate the essential characteristics of a class, according to the classical philosophy of language, which refers to Aristotle's logic. Essential characteristics, such as being a mammal or having a tail, are opposed to accidental ones, such as having an injured nose or enjoying pasta, which may occur but are not necessary in order to be classed as a member of cats.

However, even supposedly essential characteristics may occasionally be missing: a breed of cat from the Isle of Man is tailless. This has led some language psychologists, especially Eleanor Rosch, to argue that classes refer only to prototypes: although the prototypical cat has a tail, certain tailless animals can still be called cats, as they have several other characteristics of the class.

Because many documents are in the form of written texts, their words and phrases can be a starting point for KO. Nowadays words are usually recorded in a digital format, either originally or after a digitalization procedure by optical character recognition. The full text of a document can then be scanned by a search engine for any particular word or combination of words that a user is interested in. Document editors may have emphasized certain words by writing them in italic or bold type, or by putting them in the title, section headings, abstract or keyword list, and searching can exploit this by assuming that these words are especially relevant in that document.