Personal Recommendation Using a Novel Collaborative Filtering Algorithm in Customer Relationship Management

Table 1

Algorithmic performance for MovieLens dataset. The ranking score, precision, recall, intrasimilarity, and hamming distance are corresponding to 𝐿=30,40,50. And the value of parameter 𝛼 is 1.86. Each number presented in this table is obtained by averaging over five runs, each of which has an independently random division of training set and test.

Algorithms	Ranking score	Precision	Recall	Intrasimilarity	Hamming distance
𝐿=30
CF	0.148	0.077	0.321	0.316	0.704
MCF	0.131	0.087	0.360	0.304	0.751
NBI	0.120	0.089	0.379	0.291	0.778
CF-M	0.116	0.097	0.387	0.275	0.798
𝐿=40
CF	0.137	0.071	0.332	0.328	0.698
MCF	0.121	0.080	0.373	0.317	0.743
NBI	0.112	0.081	0.392	0.303	0.771
CF-M	0.107	0.088	0.401	0.286	0.790
𝐿=50
CF	0.125	0.066	0.343	0.342	0.692
MCF	0.110	0.072	0.385	0.330	0.735
NBI	0.101	0.073	0.406	0.315	0.762
CF-M	0.096	0.079	0.415	0.297	0.781