Extension-Based Recommendation System about Classification Algorithm and Its Application

The purpose of this study is to propose an approach to recommend classification algorithms for real-world classification problems. First, the extension rhombus thinking mode is used to construct performance indicators of the classification algorithm. Second, an extension-based recommendation method about classification algorithm is proposed. Third, a recommendation system is designed to implement our recommendation method. The system input is divided into two parts: expert knowledge and user preference. In order to reduce user workloads and confusions about parameter settings, a real-time interaction progressive strategy is proposed to help users complete their inputs. Additionally, the case function is designed so that users can get more support information from current and previous cases. At last, an application is used to verify the effectiveness of our system. And the application result shows that our system can provide an efficient and intelligent decision support for users.