摘要
面对爆炸式增长的信息数据,"信息过载"的问题越来越收到人们的重视,个性化推荐系统依靠其处理信息数据的优势逐渐被人们普遍接受。基于内容的推荐,基于领域的推荐以及混合推荐模型等都是当下应用非常广泛的推荐算法。在该文中,主要对推荐算法中普遍存在的数据稀疏性问题提出了针对性的改进方案,该方案有效地结合了SVD降维技术、k-means聚类算法以及相似度计算,与现存的推荐算法相比,有效的缓解了推荐系统现存的部分问题,改进后的推荐算法在准确率和误差值方面有了明显的提高。
In the face of explosive growth of information data, "information overload" of the problem received more and more attention, personalized recommendation system to rely on its advantages of processing information data is gradually accepted. Content-based recommendations, domainbased recommendations, and hybrid recommendation models are all widely recommended algorithms for current applications. In this paper, we propose a targeted improvement scheme for the sparseness of data in the recommendation algorithm. This embodiment effectively combines the SVD dimension reduction, k-meam clustering algorithm and similarity calculation. Compared with the existing recommendation algorithm, the existing problems of the recommendation system are effectively alleviated, and the improved recommendation algorithm has been improved obviously in terms of accuracy and error value.
出处
《数字技术与应用》
2017年第8期115-116,共2页
Digital Technology & Application