ABCDEFGHIJKLMNOPQRSTUVWXYZAA
1
Users and Items VariablesDatasetUser EmbeddingsItem EmbeddingsPredicted ClickError
2
IdEmbeddingsUserItemClick
3
UsersKamilla-0.17-0.43KamillaAudi Quattro1-0.167-0.432-1.0450.0000.1750.825
4
Tina0.070.48KamillaKitchen Table1-0.167-0.4320.0231.417-0.6171.617
5
Espen-0.38-0.35EspenMacbook0-0.378-0.3500.898-0.061-0.3180.318
6
Thorkild0.49-0.31EspenMacbook Pro1-0.378-0.3501.397-0.969-0.1881.188
7
ItemsAudi Quattro-1.040.00EspenAudi Quattro0-0.378-0.350-1.0450.0000.3940.394
8
Kitchen Table0.021.42TinaKitchen Table10.0700.4830.0231.4170.6860.314
9
Macbook0.90-0.06TinaMacbook Pro10.0700.4831.397-0.969-0.3711.371
10
Macbook Pro1.40-0.97ThorkildKitchen Table00.487-0.3110.0231.417-0.4300.430
11
ThorkildMacbook10.487-0.3110.898-0.0610.4560.544
12
13
Average Error0.778
14
Users and Items are represented by embeddings. In this case we are using a two-dimensional embedding (column C and D), but it could be any number.This is the dataset of the behaviour of the different users. It consist of which items the user have looked at. It also consist of the item the user have not looked at. We could have included all items the user have not looked at, but omitted that for simplicity. In column I to L we have just copied the embeddings found in column C and D for the user and the item.Our models prediction of click (1.0) or no click (0.0). Made by multiplying the user and item embeddings together and taking the sum (column I * column K + column J *column L). At the start, these numbers are completely random.The error of the model. It is simply the difference between column H and M.
15
16
17
Initial random numbers (copy+paste to start over again)
18
-0.17-0.43
19
0.070.48
20
-0.38-0.35
21
0.49-0.31
22
0.090.22
23
-0.230.32
24
0.16-0.27
25
0.29-0.45
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
Random Numbers for initialization
42
-0.10269417020.2609605542
43
0.2689827478-0.1546417604
44
0.05592306902-0.3619154042
45
0.06614384762-0.2936859784
46
0.354380528-0.4828664754
47
-0.3581916173-0.2330555745
48
0.38756849260.2313608559
49
-0.17086068030.1175077878
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100