潜在狄利克雷分配 (LDA) 主题生成

Question

Recently i have been following https://github.com/noahweber1/datacamp-project-The-Hottest-Topics-in-Machine-Learning/blob/master/notebook.ipynb to understand more on LDA. Basically it use LDA to find the hottest topic in Machine Learning from the papers.csv (NIP paper)

What confused me is the last output, the topic found via LDA.

• Topic #0 for example is for which document/row from the papers.csv?
• Are all those word in Topic is interconnected ?
• Is the word appear in all those Topic are the hottest topic or only Topic #0 is the hottest topic ?
• The topic created is not a sentence right ?

Answer 1

I have found the answer.

1. Topic #0 for example is for which document/row from the papers.csv?

Topics are just “categories”. You need to define it.

2. Are all those word in Topic is interconnected ?

Yes they are related.that's how they are generated.

3. Is the word appear in all those Topic are the hottest topic or only Topic #0 is the hottest topic ?

It will not tell you which is the Hottest topic but generally Topic #0 is the answer in this case as it related to all documents

4. The topic created is not a sentence right ?

No, the model generate the word.

More understand on the concept can be found here .