Topic modelling of TED talk transcripts
My goal is to look through the transcripts of different ted talks given by different speakers, and seperate them into distinct topics like religion, education etc. Specifically, I want to find the particular topic with the most views i.e. the most popular TED talk topic, and use topic similarity to select transcripts with the same topic.
I am using the TED - Ultimate Dataset | Kaggle created by Miguel Corral Jr and it can be found on the kaggle website @ https://www.kaggle.com/miguelcorraljr/ted-ultimate-dataset. TED is devoted to spreading powerful ideas in just about any topic. These datasets contain over 4,000 TED talks including transcripts in many languages. But I will be using only the English version.
Topic modelling was achieved using LDA (Latent Dirichlet Analysis) techniques with NLTK and Gensim libraries