Abstract: Dynamic comprehensive evaluation of hybrid information is a major research issue in comprehensive evaluation. How to effectively transform and dynamically aggregate hybrid information on the premise of avoiding information distortion has always been a hot debated and difficult point in this research area. Previous research generally presents problems such as monotonous time series data, improper transformation of hybrid information, and insufficient dynamic aggregation. To solve these problems, in this paper concentration is put on dynamic comprehensive issues with index values comprised of real number, interval number, and natural language, and an "explicit-implicit" dynamic double incentive evaluation mechanism is put forward. Firstly, to solve the problem of hybrid information transformation, two-tuple linguistic information is employed to transform natural language to ordered real number, then whole sequence method is adopted to process interval number and real number through standardization, finally a relative superiority model is proposed to statically aggregate standardized hybrid information. Secondly, LN-double incentive control line is introduced, and an "explicit-implicit" geometric double incentive model is built to dynamically aggregate time series data (static aggregated value at each time point). Dynamic comprehensive evaluation values of each plan are obtained then sorted. Thirdly, partial data from previous research is cited to analyze a vendor selection case. The case analysis results show that the proposed method is effective and superior to previous ones. The method proposed in this paper provides a new research thought and technical option for further research of similar type, including taking into consideration expert weights, collective interaction, unknown index weights or other uncertain situations.

Key words: mixed information, statically superiority model, dynamic double incentive evaluation mechanism, dynamic comprehensive evaluation