The Arabic EnConversion

The Enconverter is a language-independent parser that provides synchronously a framework for the morphological, syntactic and semantic analysis of input sentences. It is designed to perform the task of converting natural language input  into UNL format; i.e, UNL expressions, these are semantic networks made up of a set of binary relations, each binary relation is composed of two UWs and the relation they hold together. A UNL binary relation is expressed as following:

<relation> ( <uw1>, <uw2> )

The EnConverter should work with any language by simply using the language’s grammatical rules and Word Dictionary. Hence, the EnConverter should be powerful enough to deal with a variety of natural languages without being particularly inclined to a specific language family. UNL EnConversion can, therefore, support context-free languages as well as context-sensitive languages.

The EnConversion process starts off by identifying the concepts represented by the words in the input sentence, and then linking these concepts together via semantic relations to form a semantic network of binary relations; the UNL expression. There are two kinds of Arabic EnConversion rules to correspond with the two stages of analysis: the morphological analysis stage followed by building the semantic relations stage.

  1. The Morphological Analysis Stage:

In the morphological stage the EnCo grammar performs morphological analysis of words in order to:

  • Extract the correct UWs from the Dictionary.
  • Assign each UW its appropriate attributes, if needed.

Arabic morphology is non-linear and displays a wide range of inflection and derivation giving rise to a large scope of morphological variation. Unfortunately, UNL formalism is designed to segment any Natural Language input in accordance with the morphemes stored in the dictionary, this means that UNL regards any input  sentence as a sequence of morphemes; i.e., linearly. Therefore, both the inflectional and derivational aspects of Arabic morphology will have to be adapted to the linear construction of UNL.

  1. Building semantic relations stage:

After morphological analysis, the stage of assigning relations starts off, it is further divided into two sub-stages:

  • The first sub-stage aims at constructing relations between the UWs representing modifiers,
  • The second sub-stage deals with constructing relations between the UWs representing the main skeleton of the sentence.