This paper aims to find a method to detect emotion in Arabic text semantically and presents a new methodology for annotating Arabic text with an emotional category and the intensity of this emotion, in addition to the ability to address special cases when the text contains negation, words of emphasis or mitigation that affect emotion intensity, and it is the first method that can detect the implied emotion in popular proverbs in the Syrian dialect as shown in the experiments section. First, the actual need for our proposal was to overcome the absence of semantically emotion detection in both Classic and colloquial Arabic text, then a new system was developed by mapping the text with an emotional ontology and combining the semantic relations in ontology with a defined set of JAPE rules that detect the dominant emotion of the text with the intensity degree of that emotion. Our proposed system was evaluated in terms of precision and recall, it recorded 73% as precision and 72% as recall referring to our golden standard.