The Arabic parser supports equational (verbless) sentences of the form Subject – Object, and verbed sentences of one of the forms Verb – Subject – Object (VSO) or Subject – Verb – Object (SVO). Note that the parser only supports single-word objects in verbless and VSO sentences. If a multi-word object is required, you should rephrase the sentence as an SVO sentence.
Arabic verbs conjugate for mood (Active/Passive), tense (Present Indicative / Present Jussive / Past), gender (Masculine/Feminine), person (1st, 2nd, 3rd) and number (Singular, Dual, Plural). Note that in many grammar books, Present is known as Imperfect and Past is known as Perfect.
For each mood, tense and gender, the verbs list contains the following person and number combinations: 2nd person singular, 3rd person singular and 3rd person plural.
When adding new verbs to the dictionary, the parser correctly auto-conjugates all regular tri-literal verbs, and irregular tri-literal Hamzated verbs. For all other kinds of verbs, you need to manually review and edit the verb conjugations.
The Jussive form is used to form the negation of Past tense verbs. The Jussive form is also used to form uncertain sentences. Please note that in actual Arabic grammar, the form used for uncertain sentences would be Subjunctive. However, in Oracle Policy Modeling Jussive and Subjunctive are equivalent because these forms only differ in short vowels which Oracle Policy Modeling omits.
The Arabic parser supports Modern Standard Arabic (MSA). An important feature of written MSA is that short vowels (that might otherwise be denoted using diacritical marks above/below consonant symbols) are omitted and the language users infer them from the context. In particular, this means that the default Arabic verbs list does not contain these short vowels, and the parser will only recognize verbs that are written without short vowel diacritical marks. If other versions of verbs are required, these need to be added to the verbs list.
There are some Arabic verbs which begin with the Alef character and may have an implicit Hamza above Alef, or below Alef (e.g. استخدمand اكل). The default verbs list has explicit Hamzated versions of these verbs with Hamza above Alef (e.g. أستخدم and أكل), so the parser will only recognise these Hamzated versions in rule documents. If other implicit versions are required to be parsed, these need to be added to the verbs list.
In order for substitution to work, demonstratives should not be used before substituted attributes. For example, instead of the following sentence (literally "this the person which"):
مرت سنتان منذ أن تم طرد هذا الشخص من الخدمة المدنية
use "the person":
مرت سنتان منذ أن تم طرد الشخص من الخدمة المدنية
The sentence generation and substitution mechanism currently does not support attached pronouns. Specifically, when substituting a pronoun in place of an attribute, the parser replaces the attribute with an unattached pronoun. This should be correct in most cases, ie when the pronoun occurs at the beginning of the sentence. However, it may cause suboptimal sentences when the pronoun occurs at the end of a sentence. In such cases, the user may override the generated sentences as described in Customize sentence text.