I am trying to find the common topics between articles read using the respective tags attached to each article.

Background of my mini project: The problem I am trying to solve involves looking at articles read by a group of readers who have searched the same keyword, in order to gain better understanding on the nature of content they are interested in.

As I have understood, topic models are commonly used for topic extraction. I'd like some advice on whether this would be suitable for my problem, given that I already have a dataset that contains the tags ('topics') of the articles. Or would a simple probability model be more suitable?

Illustration for simple probability model:

Keyword searched: "lifestyle"

Articles read by User 1: fashion, health, organic food, clean eating

Articles read by User 2: fitness

Articles read by User 3: recipes, diet plan, clean eating

Outcome: 25% clean eating, 12.5% diet plan etc...

Sorry, I hope my explanation isn't confusing!

  • $\begingroup$ So you want to filter the users by keyword, then break down their topical distributions? If so, obviously you will need to infer the topics of the documents they read. $\endgroup$ – Emre Aug 1 '16 at 18:41

I can think of multiple approaches.

  1. You can use a simple probability model, train the model and extract the top features in order to get the nature of the interested content.
  2. As you suggested, Topic Modeling also might be useful here. You can check this out and maybe use a variant of it based on your needs.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.