I want to know about different algorithm/method/inference to extract underlying topics/themes from documents/scientific article etc.
One way would be through natural language processing or NLP. Basically, you tokenize all the words in the document and remove stop words. For the remaining words, you collect them with similar topics that you could know from dictionary learning or some other model-based technique and then calculate their term frequencies (TF). You could assign scores or weights to determine which topics it belongs to from the predefined classes, or you can use the typical classification problem of machine learning. There are a lot of related works already done in this field, for example, this could be one very simple and starting point for this. Article A practical guide to text mining with topic extraction
Hi Sandhya,
The topicmodels package is very useful and easy to use. There are lots of tutorials and examples on the web.
Regards
- Badges
- Science topic
- Similar topics
- Analytical Chemistry
- Extraction