Towards building arabic paraphrasing benchmark
Marwah Alian, Arafat Awajan, Ahmad Alhasan, Raeda Akuzhia · 2019
This research describes a paraphrasing benchmark of Arabic sentences for evaluating algorithms developed to measure Semantic Similarity for sentences and Paraphrasing identification task. The sentences are formed based on a set of rules for Arabic paraphrasing. These sentences are constructed from words of different types of Arabic books; educational, semantic science, lexicons and AWSS dataset. The benchmark consists of 1011 sentence pairs and labeled with human ratings for semantic similarity between sentence pair.