基于大数据技术的专家知识库设备画像推荐算法研究

Research on personas recommendation algorithm based on big data technologyWang Ye₁,Guo lingli₂Song Wenchao₃,,Yang shanyou₁,Cheng Long₁

摘要:

为了解决当前设备画像推荐算法采用的训练数据较少,无法充分体现设备故障特征,存在数据稀疏问题和扩展性问题的弊端,提出一种基于大数据技术的设备画像推荐算法。将专家知识库控制策略资源作为推荐资源,通过开源MapReduce技术完成对其的处理。把开源怀卡托智能分析环境和Hadoop大数据处理平台结合在一起,为设备画像推荐算法提供依据。通过k-means算法对设备故障进行聚类。利用评分矩阵相似性和设备画像特征相似度加权的思想对相似性进行计算,将相似度最大的前若干结果作为最近邻。为了保证推荐准确性,通过奇异值分解算法,针对评分矩阵通过各行设备故障评分均值对各行缺少值进行填充,获取填充评分矩阵,为目标故障推荐专家知识规则和控制策略。经验证,所提算法推荐准确性高。

Abstract:

In order to solve the problem of sparse data and scalability, a new recommendation algorithm for device based on large data technology is proposed. Large-scale multi-disciplinary devices resources are used as recommendation resources, and their processing is completed by open source MapReduce technology. The open source Waikato intelligent analysis environment and Hadoop large data processing platform are combined to provide a basis for the recommendation algorithm of device. K-means algorithm is used to device fault. The similarity is calculated by weighting the similarity of the scoring matrix and the feature of the device. The first results with the largest similarity are taken as the nearest neighbors. In order to ensure the accuracy of recommendation, singular value decomposition (SVD) algorithm is used to fill in the missing values of each row through the average of each fault score matrix, and the filled score matrix is obtained to recommend control strategies to the target devices. It is verified that the proposed algorithm has high accuracy.