I have a collection of documents. My end goal is to determine the popularity of the topic discussed by each document. In other words, whether (or to what extent) one document is discussing a popular topic. By popular, I mean the frequency of occurance of the topics.
I plan to use training topic modeling to do this task, which will yield a document-topic matrix （e.g., 1000 documents and 50 topics, i.e., a 1000*50 matrix, DoCToP). I want to use the index of the topics to represent the their popularity scores (PoPInd=[1,2,...,50], 1 being most popularity and 50 being least). By DoCToP*PoPInd, I can get a vector for each document, representing the popularity score of the topic.
I wonder is this a sound approach? Do you know any paper or project using similar approach? I tried to google myself, but little avail.
I also welcome any suggestion about how to determine the topic popularity of documents.