配电台区全业务智能检索与推荐方法研究

Research on intelligent retrieval and recommendation methods for all services in distribution station district

针对目前配电台区全业务的分类混乱、推荐准确性不高等问题,提出了一种基于KL散度和K-means的协同过滤推荐算法。采用KL变换对业务数据进行降维,减少了计算量的同时保留了数据的特征。基于KL散度对K-means算法的K值确定和初始聚类中心进行优化,提高了聚类的准确性,最后根据聚类结果,利用协同过滤算法进行业务推荐,保证了业务推荐的准确性。测试结果表明,该研究算法的性能优于其他算法。

Based on the confusion of the classification and recommendation accuracy of the entire business in the distribution station area,a collaborative filtering recommendation algorithm based on KL divergence and K-means is proposed.The KL transformation is used to reduce the dimensionality of the business data,which reduces the amount of calculation while retaining the characteristics of the data.Based on the KL divergence,the K value determination and the initial clustering center of the K-means algorithm are optimized to improve the accuracy of clustering.Finally,the collaborative filtering algorithm is used for business recommendation based on the clustering results,which ensures the accuracy of business recommendation.The test results show that the performance of this research algorithm is better than other algorithms.