How to map the weighted keywords obtained using latent dirichlet allocation algorithm (topic modeling) to a 'standard' topic-keywords CSV (that contains various topics and their associated keywords)?

See more:

Python

machine-learning

NLP

How do I map the set of weighted keywords that I got using LDA (Latent Dirichlet Allocation) algorithm (Topic Modeling) to a 'standard/known' topic-keywords CSV (this CSV contains various topics and their associated keywords)?
The output of the LDA algorithm was 10 sets of weighted keywords. Each set corresponds to a topic. The standard CSV consists of many topics and their associated keywords.
Using this CSV, I would like to get the possible topics (with probabilities) that each of these weighted keywords may have.

Any help or guidance in this is very much appreciated. Thank you!

What I have tried:

I tried training them using logistic regression (Scikit-Learn) (i.e. training on the known topics-keywords CSV and predicting the possible topics of my weighted keywords using multilabel classification), but I did not know how to take into account the weights
of these keywords (that I got from LDA algorithm) for prediction (i.e. I want the prediction to pay importance to the weights of the keywords in determining the possible topics).