医药大数据服务系统

发布时间：2018-04-30 08:51

本文选题：医药服务 + 文本相似度　；参考：《广东技术师范学院》2017年硕士论文

【摘要】：网络日益成为医药咨询、药品传递与相关交易的重要场所。但医药电商平台上的各种商品数据、销售量数据以及医疗服务数据相对独立。为整合医药电商行业产生各独立数据,发掘埋藏在数据之下潜在的有价值的信息,满足医药电商行业日益增长的大数据需求。本文给出了构建医药大数据服务系统的方案,并为医药电商提供药品推荐和销售预测服务。首先,本文介绍了医药大数据服务系统的构成。该系统是由医药大数据采集、医药数据分析平台以及报表呈现三个子系统构成。医药大数据采集子系统负责数据的采集并将数据推送给大数据分析平台。数据分析子系统负责将采集到的数据运用推荐算法和预测算法进行分析,实现药品推荐和药品销售量预测的特色功能。报表呈现子系统则负责将分析的结果以可视化的形式呈现出来。其次,本文对文本相似度计算相关技术进行研究,在对药品文本进行预处理时考虑到实际数据的特点对文本的分词结果进行特征词提取,降低了数据的维度,提高了计算结果的精确度。并结合药品说明文本特征词特点,对传统空间向量模型在文本相似度计算时无法体现特征项所具有的表现能力进行改进。实验结果表明,改进后的空间向量模型提高了药品之间的相似度以及药品推荐的精确度。最后,本文对时间序列预测模型相关技术进行研究,详细介绍了ARIMA及GM(1,1)模型。为寻求更为精确、有效的预测方法,用基于GM(1,1)模型对ARIMA模型的残差进行修正,对ARIMA模型进行优化。实验结果表明,优化后的GM-ARIMA模型达到了提高模型预测精度的目的。
[Abstract]:Network is increasingly becoming an important place for medical consultation, drug delivery and related transactions. But the various commodity data, the sales volume data and the medical service data on the pharmaceutical e-commerce platform are relatively independent. In order to integrate the pharmaceutical e-commerce industry to generate independent data, explore the potential information buried under the data, meet the growing demand of big data. This paper gives the scheme of constructing pharmaceutical big data service system, and provides drug recommendation and sales prediction service for pharmaceutical e-commerce. First of all, this paper introduces the composition of medical big data service system. The system is composed of three subsystems: medical big data collection, medical data analysis platform and report presentation. The medical big data collection subsystem is responsible for data collection and pushing the data to big data analysis platform. The data analysis subsystem is responsible for analyzing the collected data using recommendation algorithm and prediction algorithm to realize the characteristic function of drug recommendation and drug sales prediction. The report presentation subsystem is responsible for visualizing the results of the analysis. Secondly, the text similarity calculation technology is studied in this paper. In the process of drug text preprocessing, the feature word extraction of the text segmentation results is carried out considering the characteristics of the actual data, which reduces the dimension of the data. The accuracy of the calculation results is improved. Combined with the characteristics of drug description text feature words, the traditional space vector model can not reflect the expressive ability of feature items in text similarity calculation. The experimental results show that the improved spatial vector model improves the similarity between drugs and the accuracy of drug recommendation. Finally, the correlation technology of time series prediction model is studied in this paper, and the ARIMA and GM1 / 1) models are introduced in detail. In order to find a more accurate and effective prediction method, the residual error of ARIMA model is modified based on GM-1) model, and the ARIMA model is optimized. The experimental results show that the optimized GM-ARIMA model can improve the prediction accuracy of the model.
【学位授予单位】：广东技术师范学院
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.13;TP391.3

【参考文献】