分数在主题建模中表示什么

Question

I used gimsm for LSA as per this tutorial https://www.datacamp.com/community/tutorials/discovering-hidden-topics-python

and I got the following output after running it for a list of text

[(1, '-0.708*"London" + 0.296*"like" + 0.294*"go" + 0.287*"dislike" + 0.268*"great" + 0.200*"romantic" + 0.174*"stress" + 0.099*"lovely" + 0.082*"good" + -0.075*"Tower" + 0.072*"see" + 0.063*"nice" + 0.061*"amazing" + -0.053*"Palace" + 0.053*"walk" + -0.050*"Eye" + 0.046*"eat" + -0.042*"Bridge" + 0.041*"Garden" + 0.040*"Covent" + -0.040*"old" + -0.039*"visit" + 0.039*"really" + 0.035*"spend" + 0.034*"watch" + 0.034*"get" + -0.032*"Buckingham" + 0.032*"Weather" + -0.032*"Museum" + -0.032*"Westminster"')]

What does -0.708 London indicate?

Answer 1

Those are the words mostly contributing to your topic, both positively and negatively. One of the characteristics of your topic seems to be, that it does not have anything to do with London. You can see that other "London-related" words also contribute negatively to your topic: Westminster, Tower and Eye are also negative for this topic.

So if a text lacks the word London, it is highly plausible that the text is about this topic, according to your model.