一个简单的推荐系统,当作练习,在similarity package中有很多不同的相似度计算方法,包括cosine及改进、dice、欧氏及改进、jaccard、曼合顿、皮尔森、广义jaccard等方法。 在model package中主要基于模型实现矩阵分解、图推荐等的一些方法。 A simple recommendation system, as practice, has a number of different similarity calculation methods in similarity packages, including cosine and improvement, dice, Euclidean and improved, jaccard, manhattan, pearson, wide jaccard and other methods. In the model package ,including matrix decomposition and graph-based model and so on.
jiangnanboy / recomsys Goto Github PK
View Code? Open in Web Editor NEWA simple recommendation system