Abstract: This article discussed the significance of studying the semantic rules of image emotion. It gave the concept of rough set and described the learning algorithm about minimal-and-maximal rules. The author put forward the method of simplifying the semantic rules in image emotion by using this learning algorithm. Firstly, used the attribute reduction of rough set to simplify the train set. Then, got the rule set by using the decision tree algorithm. Finally, applied the lcarning algorithm about minimal-and-maximal rules to simplify the decision tree algorithm. The advantage of this method is while reducing the total quantity of the rules, it can narrow the scope of simplification, and ensure the accuracy of the semantic rules in image emotion and reduce the total number of rules.

Key words: Image emotion, Rough set, Attribute reduction, Rule simplification