Mohammed Al Logmani¹ and Husni Al Muhtaseb², ¹Saudi Aramco, Saudi Arabia and ²King Fahd University for Petroleum & Minerals, Saudi Arabia

We propose a dataset in Arabic language for automatic keyphrase extraction algorithms. Our Arabic dataset contains 400 documents along with their keyphrases. The dataset covers eighteen different categories. An evaluation using a state-of-the-art algorithm demonstrates the accuracy of our dataset is similar to that of English datasets.

Keyphrase extraction, Arabic, dataset