How to create a UW

From UNL Wiki
Revision as of 15:34, 18 February 2014 by Martins (Talk | contribs)
Jump to: navigation, search

The UNL Dictionary is never completed. It is expected to contain all the concepts that are lexicalized in at least one language. These include:

  • local concepts (i.e., concepts that are culture-bound and normally untranslatable[1];
  • local named entities (i.e., names of rivers, mountains, beaches, cities, states, neighborhoods, brands, companies, rulers, celebrities, works of art, etc.)
  • local products and practices (i.e., names of food, clothing, rituals, festivities, etc., which are specific to a given region)

All these concepts, if lexicalized (i.e., acknowledged as a "lexical unit", to be included as entries in ordinary dictionaries or encyclopedias) in at least one language, must be included in the UNL Dictionary as Universal Words.

Universal Word (UW)

A UW is a concept endowed with semantic accessibility. The semantic accessibility is granted when the concept is introduced in the UNL Knowledge Base, i.e., when we connect the concept to other existing concepts. Thereafter, the concept may be handled even by languages that do not have it yet.[2]

To include a UW in the UNLKB is to define its UCI (Uniform Concept Identifier), which is made of two parts:

  • the UCL (Uniform Concept Locator), which is a 9-digit number, automatically assigned by the machine; and
  • the UCN (Uniform Concept Name), which is an expression in the format
LRU(RELATION>CLASSIFIER)

In the above:

  • LRU stands for Lexical Realisation Unit, i.e., the name of the entity/concept. It can be a proper name (such as "Pablo Picasso", "Guernica", "Spanish Civil War", "Spanish Republican Armed Forces", "Facebook", "Candy Crush", etc.) or a common name ("paella", "baga ghanoush", "latifundium", "ilunga", etc.). For the time being, in order to ensure cross-language understanding, the name must be expressed in the way it is normally translated into English (i.e., "Spain", instead of "España", "Greece" instead of "Ελλάδα", "Egypt" instead of "مصر", "Spanish Civil War" instead of "Guerra Civil Española", etc.). Note, however, that many concepts are only transliterated into English (e.g., "paella", "baba ghanoush", "latifundium" and "ilunga" normally appear as such in English texts, even though they are not English words, i.e., they are not really translated, but borrowed, as loan words). Normally, in these cases, the words are represented in italic or between quotes in English texts, or are followed by a translator note. In any case, it is important for the LRU to be a "lexical unit", i.e., a real word (either simple or complex), and never an expression used to define the word. For instance, the LRU for "baba ghanoush" is "baba ghanoush" and not "dish of eggplant mashed and mixed with olive oil and various seasonings".
  • CLASSIFIER is a category used to disambiguate and classify the LRU. It describe a major class, such as "person", "country", "city", "brand", "
  • RELATION is a Universal Relations used to link the LRU to the CLASSIFIER. There



  • RELATION is any of the Universal Relations that can be used to link the LRU to the CLASSIFIER. S

icl and iof are Universal Relations and stand, respectively, for is-a-kind-of (icl) and is-an-instance-of (iof). The relation "icl" must be used when the concept is said to be common, where as "icl" is used when the concept is said to be proper. Compare the cases below:

  • Pablo Picasso is an instance (and not a type) of person, then: Pablo Picasso(iof>person), instead of Pablo Picasso(icl>person)
  • A painter is a type (and not an instance) of person, then: painter(icl>person), instead of painter(iof>person)
  • Metropolis(iof>city) is a specific city (the place where Superman lives)
  • metropolis(icl>city) is a type of city (a large city)


Paris is an instance (and not a type) of city, then: Paris(iof>city), instead of Paris(icl>city)

  • A metropolis is a type (and not an instance) of city, then: metropolis(icl>city), instead of metropolis(iof>city)

General Principles

UW's must comply with the following principles:

Non-redundancy
There must be no synonymy in the UNL Dictionary. Do not create UW's that have the same meaning of existing UW's. For instance: the English words "to die", “to croak”, “to decease”, “to drop dead”, “to buy the farm”, “to cash in one's chips”, “to give-up the ghost”, “to kick the bucket”, “to pass away”, “to perish”, “to snuff it”, “to pop off”, “to expire”, “to conk”, “to exit”, “to choke”, “to go” and “to pass”, when conveying the meaning of "passing from physical life and lose all bodily attributes and functions necessary to sustain life", must be represented by one single UW: "to die(icl>to change state)". The same happens to cross-language synonyms: the French words "mourir", "décéder", "périr", "s'éteindre" and "finir de vivre" must also be linked to the same UW "to die(icl>to change state)", because they convey the same meaning intended by the English words.
Non-ambiguity
UWs cannot be ambiguous. The UW is made of two parts: the UCL (Uniform Concept Locator) and the UCN (



Non-Ambiguity and Non-Redundancy A given sense may not be represented by more than one UW, and one UW may not have more than one sense. There is no homonymy, synonymy or polysemy in UNL. Arbitrariness Simple UW's are names (and not definitions) for senses. The simple UW does not bring much (or any) information about its sense. It is just a label. Any information concerning the sense is expected to be provided by the three different lexical databases available inside the UNL framework: the UNL Dictionary, the UNL Knowledge Base and the UNL Memory.

Notes

  1. Consider, for instance, the case of the word "ilunga", from Tshiluba, which means "a person who is ready to forgive any transgression a first time and then to tolerate it for a second time, but never for a third time". This is considered to be a "local" concept in the sense that it cannot be "replaced" by one single lexical item by English, although it can be "explained" in English.
  2. Consider, for instance, the case of "ilunga". "Ilunga" is a word of Tshiluba, a language spoken in the Republic of Congo. The concept conveyed by "ilunga" is not lexicalized in English or French, for instance. In this sense, "ilunga" is not directly translatable to these languages, i.e., we cannot simply replace "ilunga" by an English or French word. But this does not mean that English and French speakers cannot understand the idea conveyed by "ilunga". The only difference is that they will have to decompose the concept in several other discrete concepts (as in "person who is ready to forgive any transgression a first time and then to tolerate it for a second time, but never for a third dime"). This is the role of the UNL Knowledge Base: to interconnect concepts in order for them to be "universally" understandable.
Software