基于多层感知学习的工程档案大数据检索系统设计

Design of engineering archives big data retrieval system based on multi-layer perceptual learning

针对现有工程档案大数据检索系统检索结果不精确的问题,提出一种基于多层感知学习的工程档案大数据检索系统。该系统整体分为商用库、感知层、分类器以及实时库,并详细针对分类器型号和结构进行设计;软件中结合MLP模型对系统档案数据库中特征进行辨识,完成工程档案数据分类;设计实际使用过程中的检索序列,对系统在真正执行检索任务中工作模块调用顺序进行规划。在系统性能测试中,利用检索过程中查准率和查全率作为评价指标对设计系统和传统系统性能进行比较,实验结果表明,设计系统检索性能的综合评价结果更准确。

Aiming at the problem of inaccurate retrieval results of the existing engineering archives big data retrieval system, this paper proposes a engineering archives big data retrieval system is proposeel based on multi-layer perceptual learning. The system is divided into a commercial library, a perception layer, a classifier and a real-time library as a whole, and the classifier model and structure are designed in detail;the software combines the MLP model to identify the features in the system file database, and completes the engineering file data classification;design practical Use the retrieval sequence in the process to plan the calling sequence of the work module when the system actually executes the retrieval task. In the system performance test, the precision and recall in the retrieval process are used as evaluation indicators to compare the performance of the design system and the traditional system. The experimental results show that the comprehensive evaluation results of the retrieval performance of the design system are more accurate.