基于进化计算的现代旗袍快速推荐系统

Research on modern cheongsam fast recommendation system based on evolutionary computation

为方便用户进行个性化定制旗袍,以现代旗袍为例,对现代旗袍的部件构成和风格进行归纳与分析,选取中国风、性感风、甜美风、简约风4种典型风格的现代旗袍进行部件分类和编码,构建现代旗袍风格设计部件数据库,同时,运用交互式进化计算构建现代旗袍快速推荐系统。交互式进化计算分为2个部分:用户评价部分和种群进化部分。该系统可以结合用户需求和设计风格,在交互界面中展示满足用户需求的现代旗袍设计方案供用户选择。以实例对现代旗袍快速推荐系统进行验证,验证结果表明:运用交互式进化计算方法在选择精确度和选择时间上优于顺序查询,可以更精准快速地给用户推荐心仪的现代旗袍款式,有效缩短用户选购旗袍的时间,提高用户满意度。

Objective In personal customization of cheongsam,the existing clothing recommendation system has a single type of recommended style and the clothing style is not clearly defined,resulting in a long-time recommendation process.As such,the cheongsam matching recommendation system is difficult to meet the needs of users.Method Four typical modern cheongsam styles were selected for component classification and coding,and a database of modern cheongsam style design components was constructed.A modern cheongsam style recommendation system was constructed by using the interactive evolutionary computation method.Finally,modern cheongsam styles were displayed in the interactive interface of the modern cheongsam style recommendation system,and the main cheongsam component styles including the traditional Chinese style and sexy style were considered in the modern cheongsam style recommendation system,and the query times and query accuracy of the interactive evolutionary algorithm were verified.Results Four typical modern cheongsam styles,i.e.the traditional Chinese style,sexy style,sweet style and simple style,were selected for analysis,and a database of modern cheongsam style design components were constructed.The part library of modern cheongsam was mainly composed of style parts and decorative parts.The style part library was divided into profile,collar,sleeve,placket and slit sub-part library,and the decorative part library included fabric,color,pattern and craft sub-part library.Modern cheongsam components and styles were encoded to facilitate subsequent interactive recommendations.Interactive evolutionary computing was used to build a modern cheongsam style recommendation system,involving user evaluation and population evolution.Users were able to score the style design scheme of modern cheongsam in the interactive evolutionary computing system.The system involved individuals through the initial setting,so as to achieve the crossover and mutation operation of interactive evolution.Through evolution,a new population was generated for users to select and evaluate,and finally the evolution operation was completed,so that the system could produce the optimal modern cheongsam design scheme to meet the needs of users.The system was verified by the recommendation of traditional Chinese style and sexy style cheongsam.The verification results indicated high accuracy and high efficiency of proposed method.Conclusion The research uses the interactive evolutionary algorithm to build a modern cheongsam style recommendation system,and the system is mainly based on the modern cheongsam style components classification,scientific analysis.The modern cheongsam components are encoded,which is conducive to the recommendation system to efficiently pass information to users.The modern cheongsam recommendation system uses the interactive evolutionary calculation method to be superior to the sequential query in the selection of intensive reading and selection time.It can recommend the modern cheongsam style to the user more accurately and quickly,and can effectively improve the user′s satisfaction with purchasing modern cheongsam.The modern cheongsam recommendation system can quickly grasp the user′s preference for modern cheongsam style,greatly reduce the time for users to select cheongsam,make users more efficient and convenient to find personalized and customized modern cheongsam,and have innovative significance for modern cheongsam interactive style recommendation.In the future,the modern cheongsam recommendation system can also establish a variety of style library,to bring users personalized custom style of modern cheongsam recommendation system for the field of interactive clothing to provide reference and reference.