nlp - Best model for topic spotting/discovery -
what best model topic spotting within short unstructured documents, ex. sms or twitter messages? latent dirichlet allocation?
lda 1 of strongest models available topic modeling, applying short texts such twitter/microblog posts might require work. authors of this paper discuss lda , alternative model , recommend aggregating multiple posts before running topic model on it.
[watch out terminology: "topic spotting" old synonym supervised document classification.]
Comments
Post a Comment