Arabic language is one of the popular language that has a tremendous types of nested noun compounds. Nested noun compound refers to the combination of two or more nouns or nouns with adjectives. Recently, different research studies have addressed the process of extracting these NCs. However, most of the literature has focused on statistical measures which may lead to inaccurate extraction. This study aims to address the capability of machine learning for extracting Arabic NCs. The proposed method consists of a feature extraction phase where different features are being generated. Consequentially, a Naive Bayes classifier has been trained on such feature extracted representation. Experiments have been conducted on a benchmark Arabic corpus. Results showed that the proposed method has obtained an average precision of 0.86. Comparing such result against the state of the art reveals that machine learning could outperform the traditional statistical measure in terms of extracting NCs.
Volume 10 | Issue 12
Pages: 478-485