Thesaurus

While some works called "thesauri", such as Roget's Thesaurus, group words in a hierarchical hypernymic taxonomy of concepts, others are organised alphabetically[4][2] or in some other way.

[6] The word "thesaurus" comes from Latin thēsaurus, which in turn comes from Greek θησαυρός (thēsauros) 'treasure, treasury, storehouse'.

[20] Roget described his thesaurus in the foreword to the first edition:[21] It is now nearly fifty years since I first projected a system of verbal classification similar to that on which the present work is founded.

Conceiving that such a compilation might help to supply my deficiencies, I had, in the year 1805, completed a classed catalogue of words on a small scale, but on the same principle, and nearly in the same form, as the Thesaurus now published.

The book starts with a Tabular Synopsis of Categories laying out the hierarchy,[23] then the main body of the thesaurus listed by the Head, and then an alphabetical index listing the different Heads under which a word may be found: Liable, subject to, 177; debt, 806; duty, 926.

[27] In addition to its taxonomic organization, the Historical Thesaurus of English (2009) includes the date when each word came to have a given meaning.

For example, three different senses of "debt" are listed in three different places in the taxonomy:[28] A sum of money that is owed or due; a liability or obligation to pay An immaterial debt; is an obligation to do something An offence requiring expiation (figurative, Biblical) Other thesauri and synonym dictionaries are organized alphabetically.

Francis March's Thesaurus Dictionary gives for liability: CONTINGENCY, CREDIT–DEBT, DUTY–DERELICTION, LIBERTY–SUBJECTION, MONEY, each of which is a conceptual heading.

[11] Before Roget, most thesauri and dictionary synonym notes included discussions of the differences among near-synonyms, as do some modern ones.

A thesaurus can form part of an ontology and be represented in the Simple Knowledge Organization System (SKOS).

[41] Thesauri are used in natural language processing for word-sense disambiguation[42] and text simplification for machine translation systems.