Can CTM return topic_significance_uniform score?

[inceatakan]

• OCTIS version:
• Python version:
• Operating System:

Description

Currently, CTM returns only topic_significance_background score. Is it possible to get topic_significance_uniform score (per topic)?

Thanks,
-Atakan

What I Did

Paste the command(s) you ran and the output.
If there was a crash, please include the traceback here.

[silviatti]

Hi, what do you mean by "CTM returns only topic_significance_background score"? Can you show a code snippet?
Once you get the model output, you can use it to compute every other evaluation metric that you like. For example:

from octis.evaluation_metrics.significance_metrics import KL_uniform
from octis.evaluation_metrics.significance_metrics import KL_background

kl_u = KL_uniform() # Initialize metric
kl_b = KL_background() # Initialize metric

model = CTM(num_topics=25)  
model_output = model.train_model(dataset) 
kl_b_score = kl_b.score(model_output) # Compute KL-background score
kl_u_score = kl_u.score(model_output,per_topic=true) # Compute KL-uniform score

Hope this helped :)