CAJ | 학술논문

针对目前图书采购过程中经验性、主观性强，统一性、科学性差等缺点，提出一种基于遗传神经网络用于图书采购的建模方法。该方法首先引入遗传因子改进传统遗传算法，并以此对神经网络的权值、阚值及其结构进行优化，然后通过优化后的神经网络挖掘图书的各种属性与是否被采购之间的潜在关系，从而实现图书是否被采购的预测分类。仿真实验表明该图书采购模型具有良好的预测性能及泛化能力，具有推广价值。
침대목전도서채구과정중경험성、주관성강，통일성、과학성차등결점，제출일충기우유전신경망락용우도서채구적건모방법。해방법수선인입유전인자개진전통유전산법，병이차대신경망락적권치、감치급기결구진행우화，연후통과우화후적신경망락알굴도서적각충속성여시부피채구지간적잠재관계，종이실현도서시부피채구적예측분류。방진실험표명해도서채구모형구유량호적예측성능급범화능력，구유추엄개치。
Aiming the disadvantages in the process of book purchase, a book purchase model based on genetic algorithm neural network was proposed. Firstly, we improve the performance of genetic algorithm through influence factor and optimize weights, thresholds and structure of neural network simultaneity with the improved genetic algorithm. Then the optimized neural network was used to mine the potential relation between books＇ attributes and whether will be bought. The simulation experiments show that the book purchase model has a good capability of forecasting and generalizing.