Chapter 2  UWs of UNL

There are five kinds of UWs: Basic, Restricted, Extra, Compound and Temporary UWs. The first three kinds of UWs are the real UWs which should be stored in the UW Dictionary and linked with each other in the UNL Knowledge Base (KB). A Compound UW is not a simple UW but a scope. Temporary UWs do not exist neither in the UW dictionary nor in the KB directly, they only appear in the UNL documents.

This chapter explains what are the Temporary UWs and NULL UWs in detail, and how each kind of the UWs should be used. For details of the Basic UWs, the Restricted UWs and the Extra UWs, see the UW Manual. 

2.1 Basic UW 

The Basic UWs are character strings that correspond to English words. Each of them represents all the concepts that the English word covers. For details, see the UW Manual.

When a Basic UW can be used?

For instance, the UW represented by “book” can express the both meanings of “a book” and “to book”. Such a Basic UW can be used in the UNL expressions of a sentence. If it is used, which meaning it expresses must be possible to distinguish by the relationships with other UWs included in the sentence. Otherwise, it is not recommended to use such a basic UW. 

2.2 Restricted UW 

The Restricted UWs are defined by attaching a Constraint List to the Basic UWs. It represents a subset of the concept of the Basic UW, a more specific concept. For details, see the UW Manusl.

For instance in case of “book”, the meaning of “to book” can be expressed as “book(icl>do)”, and the other can be expressed as “book(icl>thing)”. These “book(icl>do)” and “book(icl>thing)” are Restricted UWs. 

2.3 Extra UW

The Extra UWs are defined by the words of native languages excluding English using the Roman Alphabet. The Extra UW is very useful for defining an UW for a thing that is native to a special language. If there is no a correspondent word in English for defining the UW for such a language-native concept, the Extra UWs provide the way to define it.

As for an Extra UW, it is recommented to use restrcitions in the UW definition although its meaning is unique. For instance, “tatami(icl>thing)” can be used to define the UW for the word of “tatami” in Japanese. 

2.4 Compound UW

A Compound UW indicates a set of binary relations of UNL. It is also called a Scope. A scope is useful in linking a group of UWs to other UWs in the UNL expressions. Such a Compound UW can be treated as a symple node. The details of a scope are explained in "3.3 scope". 

2.5 Temporary UW

There are four kinds of Temporary UWs in UNL. The Temporary UWs only appear in the UNL documents but neither the UW Dictionary nor the KB. They are the following.

First one is the Arabic Numerals. A number in UNL is expressed by a Arabic Numerals. For instance, the UW for the number “five thousand” is 5000. On the other hand, an Arabic Numerals can also be a nominal concept as in “0 is the smallest number in natural numbers”. To treat a Arabic Numerals in an UNL expression as a number or a nominal concept, it depends on the relationship between it and the other UWs. For instance, when “qua” or “bas”is pointing/entering into an Arabic Numerals, it then should be treated as expressing the concept of a number. An English word like “five” can also appear in the UNL expressions, in this case the UW “five” only expresses the meaning of the nominal concept but not a number.

Second one is the Ordinal Numbers. An Ordinal Number in UNL is expressed by attaching the attribute @order to the Arabic Numerals. For example, the UW for the concept of ordinal number “second” will be 2.@order. The same the English word “second” can also appears in the UNL expressions for expressing the meaning of the nominal concept but not a number.

Thrid kind is the Labels, Symbols, Formulas, etc. It is allowed to express a Label, a Symbol or a Formula in the UNL expressions in Double Quotes, as “1.1”, “http://www.unl.ias.unu.edu” and “agt(eat.@entry, I)”.

The last one is the Unknown UW. An Unknown UW is also quoted by Double Quotes in the UNL expressions, as “Unknown UW”. There is almost no difference between an Unknown UW and a Label, except for an Unknown UW is mainly generated by the enconversion system when it fails in finding an appropriate word from the dictionary.  

2.6 Null UW

A NULL UW is allowed in the UNL expressions by using a null string as "". The null UW means any concept that can be replaced by other UW. It is useful in making the UNL expression of an elliptical sentence.

For instance, in the sentence "I'm happy if you are", the second "happy"  is thought to be omitted normally as its full sentence should be "I'm happy if you are happy". The following UNL1 and UNL2 show the UNL expressions when the UW for the omitted word “happy” is filled up and when the null UW is used respectively.

Example 1

In the above example it is clear that what is omitted but may not be always. For instance, in case of the following example "talk over a glass of beer". In this sentence, no word is omitted syntactically, but from the point of view of an UNL expression, a verbal concept like "drink" or "have" is necessary for expressing the its meaning clearly. It is fully correct to find the corresponding word by means of referring to the co-occurrence dictionary for example if possible, but it is not always guaranteed.

The null UW is introduced for substituting such UW that is necessary in an UNL expression but its word does not appear in the original sentence and there is no way or it is difficult to find it out. The following shows another example in which an null UW is used in the UNL expressions.

Example 2

How to distinguish a scope node and a null UW?

In case of a null UW, the null string expression of “” is always necessary. When more than two null UWs are necessary in a sentence, different identifiers are given to each of them, as in "":01 and "":02.

A scope node always begins with the colon “:”.

Is it possible to omit the NULL UW expression ""?

The null string of "" for an NULL UW cannot be omitted.