A Possibilistic Approach for the Automatic Morphological Disambiguation of Arabic Texts

A hybrid possibilistic approach for Arabic full morphological disambiguation

“…Some features are binary and the value which begins with the letter 'N' means that this word does not contain this feature. However, POS, pronoun, person, voice, aspect, gender, number, case and mode have more than two possible values [23,24].…”

“…Arabic text disambiguation consists in classifying the different morphological features through a perfect or an imperfect training set [23,24]. Thus, we study classification approaches to disambiguate non-vocalized texts using vocalized texts for training.…”

“…The Max operator is used to get normalized frequencies [37][38][39]. In the imperfect case, we add the imprecision rate β jk of the attribute A j in the instance I k [24]. Hence, Occ(a jL, c i ) turns into:…”

“…We use Arabic vocalized texts as input data. For each word of the text, we compute the values of 14 morphological features (MF) as in [23,24]. Therefore, we use an extended version of the morphological analyzer AraMorph [59] that handles vocalized and non-vocalized texts.…”

“…The ambiguous instances pass through the possibilistic classifier to determine the accurate value. To do so, we compute possibility and necessity measures over the training sets [23,24] and we include the information gain of each morphological feature. Referring to the measures previously calculated, we determine the best class value for each morphological feature (cf.…”