In recent years, there has been increased interest in eventdetection using data posted to social media sites. Automaticallytransforming user-generated content into informationrelating to events is a challenging task due to the short informallanguage used within the content and the variety oftopics discussed on social media. Recent advances in detectingreal-world events in English and other languages havebeen published. However, the detection of events in the Arabiclanguage has been limited to date. To address this task, wepresent an end-to-end event detection framework which comprisessix main components: data collection, pre-processing,classification, feature selection, topic clustering and summarization.Large-scale experiments over millions of ArabicTwitter messages show the effectiveness of our approach fordetecting real-world event content from Twitter posts.