ABCDEFGHIJKLMNOPQRSTUVWXYZ
1
2
Welcome to the Guided Topics Visualization Template!
3
This Google Sheet was made to help you quickly understand the quality of your seeded topics and other topics discovered by Guided Topic-Noise Model.
4
5
Using this Sheet is simple!
6
Step 1: Copy your seed topics into the Seed Topics tab. Each topic should be its own column.
7
Step 2: Copy the topics generated by GTM into the Guided Topics tab. Each topic should be its own column.
8
9
After Steps 1 & 2 are completed, you should see seed words from the Seed Topics tab get highlighted in the Guided Topics tab with the color corresponding to their topic.
10
If you see columns with a bunch of words with the same color, this indicates that GTM has done a good job finding a seed topic.
11
12
If you see columns with a bunch of different colored cells, or few or no colored cells at all, this means that GTM did a bad job finding a seed topic. There are a couple common ways for this to manifest in the visualization:
13
1. The seed topic does not exist in the data set. This is common when there's just not enough data, or when the seed topic really doesn't belong in the topic set. This is easy to identify when there are few or no colored cells with the right color for the seed topic.
14
2. Two or more seed topics should be combined into one. Often, we might start out believing that there are two separate topics, only to find out that they are two sides of the same coin. This is easy to identify when there are a bunch of colored cells, but only two major colors (two topics combining into one).
15
3. One seed topic should be divided into two or more topics. If we see that a seed topic was found, but it diverges from our intended seed topic, it might be the case that we can divide the seed topic into multiple seed topics, resulting in smaller but more coherent sub-topics. This is harder to identify, but if you see that only a small subset of your seed words for a topic are appearing, along with a bunch of other related words, it is worth splitting those seed words into their own topic for the next iteration.
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
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