التحليل الاحتمالي للعبارات الثنائية الحدين في العربية : دراسة مدونية حاسوبية

الملخص EN

International research using experimental, intuition-based, and corpus-based methods indicates that there are factors can play a role in the ordering preferences of binomials.

in this paper, i analysed 1000 Arabic binomials based on the sub-corpus of (Arabic Web 2012 : arTenTen12) which is consisted of 429,990,596 tokens from Standard Arabic texts grammatically tagged with stanford tagger.

the paper has experimentally found through conducting a logistic regression analysis of binomials that five constraints mentioned in the previous literature affect binomial ordering.

the statistical and probability values of our model showed a variation in the ranks of influence of these constrains on the binomial ordering.

it was found that the most influential is markedness, while alphabetic was excluded due to statistically non-significant.

the best model with the four constraints was able to explain about 0.48 of the fixedness of binomial, and it also revealed learnability and showed good results in the measures of performance.

it is expected that increasing the size of the data and adding other constraints will increase contributing to the interpretation of the binomial ordering and will improve model performance.